Recent posts in Interview Experience

238
238 views

Many candidates who have appeared for NIC Interview have shared valuable insights from their personal experiences, giving future aspirants a better understanding of what to expect on the interview day. These firsthand experiences can be extremely helpful during interview preparation, as they highlight the key topics to focus on, commonly asked questions, and recurring interview trends.

YearInterview Experience Blog LinksSpecialization
2022Pushpraj's ExperienceCSE
2021Sayan's ExperienceCSE
2021Pawan's ExperienceCSE
2018Saswat's ExperienceCSE

 

See ALL our interview experiences arranged institute-wise and year-wise HERE.

To share your interview experiences: Add your NIC Interview Experience

1,140
1,140 views

Hi Everyone,

I am Jayesh. I wrote GATE CS in 2026 and secured AIR 85 with score of 899.

For this program, the minimum GATE score required for consideration was 650. However, since the number of applicants was quite high, a notification was posted on their site indicated that the last candidate shortlisted for the programming test had a GATE score of around 713.

Procedure: -

  1. Firstly, one had to apply for this program on IIT Bombay portal within stipulated deadline. This year it was 10th April.
  2. Then around 22nd April, the intial shortlist was released on their website. Simulatenously, a Google Form was circulated asking for project preferences. My top preference was System Adminsitration. It was the only position I was interested in. But they ask for top three preferences. Apart from preferences, they also ask for your most impactful project during Bachelors, link to your programming portfolio and a research paper that you have enjoyed reading. Everything except preferences was optional to fill.
  3. After submitting the form, candidates had to appear for an onsite programming test. Sample questions were shared on the website beforehand. This year, the test was conducted on 12th May and consisted of four programming questions. It was held on IIT Bombay's proprietary platform, BodhiTree (similar to Codeforces). The initial duration was 2 hours 15 minutes, which was later extended by 45 minutes. Based on my estimate, solving at least 2 questions was enough to qualify for the interview. However it was mentioned that criteria for being shortlisted for interview was 70% GATE Score + 30% Programming Test. This round was purely eliminatory and had no effect on final selection.
  4. Those who had cleared the programming test on 12th May were informed in the evening and had to appear for interview on 13th May. Interviews started from around 9: 30 AM. Students based on their performance were at most selected for 2 interviews.

Now to the questions in programming test: -

An Ubuntu desktop with access to the BodhiTree platform was provided. All compilation and debugging had to be done through the terminal. VS Code and the C++ reference manual were also available. Only C++ was allowed.


1) Given a number $n \in [1, 13]$. Print the numeric pattern shown as below: -

Example: 

n = 501 02 04 07 11

03 05 08 12

06 09 13

10 14

15

 

2) Given an integer $n$ and an $n \times n$ matrix, create a $2n \times 2n$ matrix by placing the original matrix and its flips in all four quadrants.

Example:

3 1

2 5
3 1 1 3

2 5 5 2


2 5 5 2

3 1 1 3

(Top-left: original, top-right: horizontal flip, bottom-left: vertical flip, bottom-right: both flips.)

3) Given a postfix expression, compute its value. The problem itself was straightforward; the main challenge was parsing the input string correctly and tokenizing it into integers and operators.

Example:

5 1 2 + 4 * + 3 -14
 
4) This was more of a comprehension exercise than a coding problem. You had to read the provided PDF to understand the task; the following document and starter code were supplied: https://www.indiapost.gov.in/documents/offerings/intiatives/DIGIPIN_Technical_document.pdf
In short the problem was: given a latitude and longitude, convert them into a DIGIPIN code. Starter code and helper functions were included; your job was only to implement the function that returns the DIGIPIN (a 10-character string). The main challenge was understanding the specification in the PDF and the starter code that was given, once that was clear, the programming part was straightforward(not more than 40 LOC).
 
Example: 
Lat: 19.1335144659

Long: 72.9092113631
DIGIPIN: 4FK26P2L67

I was able to solve all four questions. Based on which, I was shortlisted for the System Administrator interview (my 1st preference) and the EdTech project (my 3rd preference) led by Prof. Kameshwari Chebrolu.

Approximately 25–30 students were shortlisted for the System Administration position, 55–60 for the Web Development position, 20–25 for Prof. Kameshwari Chebrolu's project, and 40–50 for the CS 101 position.

In the interview list they shared, I was the 1st candidate to be interviewed for System Administration position and the 2nd candidate for the EdTech project. The reporting time was 9:30 AM; however, I reached the venue around 10:00 AM. Had I arrived on time, I could have finished both interviews by around 12:00 PM. But since I reached late, my interview for Prof. Kameshwari Chebrolu's project was conducted first at around 11:00 AM, followed by the System Administrator interview after the lunch break at around 3:00 PM.

Prof. Kameshwari's EdTech Project Interview: -

The panel consisted of Prof. Kameshwari Chebrolu and three M.Tech. RAs.

I felt this interview was fairly generic. It didn't go into much technical depth maybe because I stumbled on basic questions or they were mainly trying to understand who I was and what kind of work I had done. They mentioned a few common technologies and asked questions around them. Some of the questions I remember are: -

  1. They asked about Data Warehouses and Databases, specifically the difference between OLTP and OLAP, since I had mentioned them in my project description in the Google Form. I explained the differences briefly.

  2. They then asked which type of database would be suitable for implementing the chat feature in an app like Instagram. I answered a relational database because of the implicit relationships in messaging applications. Looking back, I think this was where I butchured my interview.

  3. Since this was an EdTech project, Prof. Kameshwari asked how I would design a multithreading lab where the system could verify whether a student had actually created multiple processes. I had no idea how to answer this.

  4. Finally, she asked whether I was familiar with the ss utility in Linux and how I would verify whether a student had used it. Fortunately, I was familiar with it through the OverTheWire Bandit exercises, so I was able to answer this question. She seemed satisfied.

Overall, I don't think I performed well in this interview. Partly because I wasn't particularly interested in this project, and partly because I had prepared almost exclusively for the System Administration interview. However, in hindsight, it was beneficial that this interview happened first, as it helped me identify a few things I needed to improve before my main interview.

The interview lasted around 10 minutes. Some other candidates who felt they performed good had their interviews duration for around 20–25 minutes.

System Adminstration Interview: -

The panel consisted of Prof. Soumen Chakraborty and 5 M.Tech. RAs.

They began by asking my introduction. I briefly talked about my academic background, work experience, GATE score, and, since this interview was focused on Linux, I also mentioned that I had credited a Linux course during my bachelor's degree and that Linux is my primary operating system.

They then asked which Linux distribution I use. I replied that I use Omarchy, an Arch-based distribution, while also mentioning a few other distributions I have used. After that, they moved directly to Linux-related questions. All Linux based questions were asked by RAs only. Few questions I remember were: -

1. What happens when we type a command in the shell?

I started by explaining globbing and briefly described how the shell expands wildcard patterns. I then mentioned that the shell searches for the command's executable.

2. How is the command binary located?

I answered that the shell searches for the executable in the directories listed in the $PATH environment variable. Once the executable is found, a new process is created and the command is executed with the supplied command-line arguments.

3. They weren't fully satisfied and followed up by asking about the exec keyword.

At that point, I realized they were referring to the exec and fork system calls involved in process creation. I explained the role of both in the process lifecycle, after which they seemed satisfied and moved on.

4. What is the difference between an absolute path and a relative path in Linux?

I briefly explained the Linux directory hierarchy, starting from the root directory (/), and then described the difference between absolute and relative paths.

5. Given an absolute path and a relative path, how would you convert the absolute path into a relative path with respect to the given path?

I explained my approach using an example. I said I would model the directory hierarchy as a tree (or Trie-like structure) to determine the common path and construct the relative path.

6. Given two absolute paths, how would you find their least common ancestor?

I again described a tree-based approach to identify the deepest common directory. They seemed satisfied with the explanation.

7. They asked whether I knew about Linux security. I admitted that I wasn't very familiar with it. They then switched to networking and asked how a packet travels across the Internet.

I explained the process starting with DNS resolution (checking local cache and global DNS lookup), followed by the establishment of a TCP connection (for TCP traffic), packet forwarding through routers, and finally reaching the destination host.

8. They still weren't satisfied and asked how a packet from a private network reaches the public Internet.

I answered that this is done using Network Address Translation (NAT). They seemed satisfied with this answer.

9. How would you find the five largest files in a directory?

I replied that there was probably a single command for this, but I couldn't recall it at the time. Instead, I suggested using a pipeline of commands: ls -la to list files, cut to extract the relevant fields, sort to sort by file size, and head -5 to obtain the largest  5 entries.

10. They then asked whether ls -la reports directory sizes correctly.

I honestly replied that I didn't know.

11. Which command would you use to obtain the size of a directory?

I immediately answered du.

When I mentioned the Trie data structure, it seemed to catch the professor's attention. From that point onward, the all questions were asked directly by the professor rather than the M.Tech. RAs.

The following questions were asked: -

1. Given a collection of strings, how would you find their longest common prefix?

I initially suggested using a Trie. While inserting each string, I proposed maintaining a count at every node to keep track of how many strings passed through that node.

2. Is maintaining the count necessary?

After thinking for a moment, I realized it wasn't. As long as the Trie does not branch into two or more children, all strings share the same prefix up to that point. The longest common prefix is simply the path until the first branching node.

3. Which subjects would you be interested in pursuing during your Master's?

I mentioned that I am more inclined towards the systems side of Computer Science, so I would be interested in subjects such as Computer Architecture, Operating Systems, and related courses.

4. What exactly in Computer Architecture interests you?

Initially, I wasn't sure how to answer. I explained that my undergraduate coursework covered topics up to pipelined processors and in-order execution. I said I would like to learn more about superscalar processors and out-of-order execution. He seemed satisfied with this answer.

5. Given two arrays a & b of size m & n respectively, compute: $\min_{1 \le i \le m,; 1 \le j \le n} |a_i - b_j|$

At first, I couldn't understand the problem from its verbal description and spent a couple of minutes clarifying it. Eventually, he wrote the expression on his iPad and asked me to describe a brute-force solution.

I proposed using two nested loops to evaluate ($|a_i - b_j|$) for every possible pair while maintaining the minimum value encountered. When asked about the time complexity, I answered $O(n*m)$.

He then asked me to optimize the solution. After taking a couple of minutes to think, I proposed sorting one of the arrays and, for each element of the other array, using binary search to find the closest element in the sorted array. This would reduce the complexity to $O((m+ n)\log m)$.

He then asked which array should be sorted. Initially, I had said the larger array, but after analyzing the time complexity, I corrected myself and said it would be better to sort the smaller array.

He remarked that this was asymptotically optimal but then asked if I could avoid performing a binary search for every iteration.

After thinking for a while, I suggested sorting both arrays and using the two-pointer technique. Starting with pointers at the beginning of both arrays, we first compute ($|a_0 - b_0|$), update the minimum, and advance the pointer corresponding to the comparision between $|a_1 - b_0|$ & $|a_0 - b_1|$. 

They seemed satisfied and by this point, around 30–35 minutes had passed. They told me the interview was over, and I left. Overall, I felt the interview went well.

Note:-

1) All cutoffs I mentioned are for Unreserved category.

2) Apart from your preferences, everything was optional to submit. Your preferences are extremely important. Please fill the form on time and carefully. There were few students who forgot the fill the form. They weren’t even called for programming test.

3) For programming test, you should know how to compile and run the executable via CLI.

4) This year, Interviews started at 9:30 AM, and they informed at 9:00 AM. Better to be prepared in advance. I was preparing for the interview till the last moment. So, basically reached late.

5) Finally, the result of this program are directly released on COAP atleast for this year and on round 4. If you’ve received an offer in Round 1, you may have to let that go because offer for this program in later rounds(this year round 4). Same happened with me. I didn't come to know the final result as I had already accepted the offer in Round 3 which I couldn't retain for any further rounds.

6) Resources that helped me prepare for Linux interview: -

 

 

 

709
709 views

Entered the interview room around 1 pm, I was one of the last candidates to be interviewed among first half slot of candidates on 18th July (first day of interviews for M.Tech CS/CrS).
It was a medium sized classroom with 6-7 professors (don't rememmber exactly how many were there). 

They asked me to first take a seat and told me to Introduce myself briefly.

Then a professor ask, What are the topics you like?
Answered: Graph Theory, Set Thery and Combinatorics in Mathematics, TOC, DS in CS. 
Algorithms? They asked. I said I like the subject, but not confident in it. 
Do you like coding? Nodded with an yes.

Starts with Graph theory.

  • Define a graph.
    Answered by writting the definition on the whiteboard.
  • Okay can Draw a graph. Can you draw another edge between 2 vertices(which were already connected). Then what would you call that, does your definition suffice for this graph?
    I got little confused with the mathematical notations for directed, undirected, simple and multi graph
  • How many different possible combinations of edges can a graph have?
    I answered by starting with maximum bound for edges and then showing for every possible pair of vertices, an edge may be or may not be present, giving 2 choices. He further asked what would be the count if the graph is directed, answered that as well.

Next they move to counting.

  • He tells me to define a function from a set of n elements to itself. How many possible mappings are possible?
    Answered.
  • How many bijections are possible?
    Answered. (Although this seemed like a trivial question for me, I still explained my process before reaching the final answer)
  • Now he said to write the elements of A as $1,2,...,n$. Gives the definition of inversion: $(1,2)$ an inversion pair if $f(1)>f(2)$. Then he asks me to calculate how many such functions are possible where the pair $(1,2)$ is inverted.
    Took me a while to understand what we meant by inversion, tried to relate it to inversion pairs in sorting. Worked it out, and made a small mistake, he told me to explain my working, and I was able to figure out and correct my mistake. At the end I was able to answer correctly.

Next they wanted to ask from Linear algebra.

I was hesitant about this topic as I hadn't revised it. So ma'am asked from general algebra specifically Polynomial. 

  • Write a n-degree polynomial. It has a special property i.e. $F(x^2)=(F(x))^2$. Then what can be said about this polynomial?
    Polynomials was a new topic for me, so it took me time to understand what is happening. With professors' help I was able to compute what will be the coefficients $c_0$ and $c_{2n}$. But I wasn't able to calculate for general $c_i$ which would give me the complete answer. if I had more time I might have solved it, they said it's okay you can try to solve it later at home.
The interview went for around 25-30 minutes. I felt my interview was okayish, and I knew my PCB did not go that well. My PCA section on the other hand went pretty well.
For those who do not know, ISI Entrance Test has 2 sections: PCA (MCQ with questions mostly from maths, and some CS areas like DSA, Algorithms, TOC), PCB (There are 2 group and one has to answer either one of them, one is CS and other is Non-CS, the theoretical CS questions were common to both groups)
 
Final result: Selected for M.Tech Cryptology and Security in Round 1. General Merit List rank of 14.
For context:
  • My Gate score was 629, not something I am proud of, as I have been preparing for 2+ years (with college).
  • I felt more comfortable with ISI PCA questions, and JEST TCS questions as opposed to GATE questions.
  • I have solved some previous year questions of TIFR, ISI and, CMI from respective GO PDF. I haven't prepared mathematics separately for ISI Entrance Test.
 
Thank you to GATEOVERFLOW platform for helpful solutions and discussions of different previous year questions of GATE, ISI, CMI and TIFR exams.
335
335 views

At the start, they asked about my 10th and 12th math percentages, my CGPA, and my GATE score.

Then they asked why my PCB score was good but not my PCA. I answered honestly, saying that my 10th and 12th-grade math is not very strong, but I am very good at core CS subjects.

Then they asked my favorite math topics and CS subjects

I said for math topics Graph Theory,Probability For CS : DSA , DBMS

Question 1: Given an n-vertex simple graph, prove that all vertices cannot have distinct degrees. Answer: Proved using the Pigeonhole Principle.

Question 2: Given an n-vertex graph where each vertex has a degree of 2, prove that it has a cycle. Answer: I was stuck here initially, but based on the professors' hints, I answered by proving the graph has exactly n edges (which is > n-1 edges) using the Handshaking Lemma.

Question 3: How do you find a cycle in a graph, and what is the time complexity of the algorithm? Answer: Using DFS (Depth First Search), and I explained why the TC is O(V+E).

Question 4: Given n distinct items, if you have to pick 4 items with repetition allowed, what is the probability of selecting 4 distinct items? Answer: [n * (n-1) * (n-2) * (n-3)] / n^4

The professors were very friendly and helpful throughout the entire process.

Final Result : Selected for MTech CS

272
272 views

 

I used to read a lot of blogs on GO website during my GATE prep and always thought i will write one one day so here we go,

A little bit about me ,I am a dropper graduated in 2025 and f*cked up GATE on the day of the exam. So i immediately started preparing for PGEE , I had originally planned to apply for MS instead of Mtech in IIITH but applied for Mtech as my gate score was really low and i knew i woudnt get any good clg with it  and didn't want to be empty handed in the end(dint want to take another drop) and there was no guarantee that i will get MS because it solely depends on the interview performance so i didn't want to take any risk and applied for Mtech. As fate would have it ,I was short of 1 mark in PGEE and didn't get Mtech seat also but thankfully I got a chance to apply for MS and got shortlisted for interview.

on the day of INTERVIEW

My interview was scheduled on 10th june . I arrived a little late around 9 am, as my train was late. I had already started panicking on my way because they told to report at 8 am. Once I reached the hall there were several volunteers present . After verification I saw that I had been shortlisted for CVIT and LTRC my first two preferences and the volunteers told me to go have breakfast as there was still time left before the interviews began 
 

LTRC

I was first led to LTRC along with 7-8 ppl. This interview was vey short lol. I was a blaberring stuttering mess the whole 10-15mins it lasted. I was very nervous and anxious. They verified my name and directly started asking questions.There were 3 profs present. They didn't ask for my resume or SOP.

P1: What subjects have u prepared 

Me: ML,LA 

P1: ok what do u know in ML, do u know classical ML algorithms

me: yes sir

P1: what is supervised and unsupervised learning

me: gave the genral definition 

P1: what do you mean by labels

me:proceeded with describing features as labels

this is where things started going south imo, I couldn't get the right words and was blaberring. another prof P3 took over and kept asking me q untill I realised my mistakes.then P1 asked about clustering algorithms I answered corectly this time. Then P2 took over

P2: asked me what was my mother tongue

me: Tamil

P2: lets say I have one very big stack of tamil newspapers and i want you to build a dictionary using the words available in the papers .how will you do

me: first we should tokenize sir then we can start building the dictionary

P2: what does tokenization mean

me: I answered

P2:how will you know on what basis to create the tokens for example your name Srividhya is one meaningful word but Sri and vidhya separately are also meaningful words.

me:we can use some algorithms which can use Machine Learning to find such words.(I dint know the answer and answered what came to my mind).
 
 P2 then asked if anyone else wanted to ask q P1,P3 said no and they asked me to leave.
 
I had a project where I had used transformers and was expecting questions on that so i focussed on LLM's mainly during my preparation and didn't bother with other topics but the panel didn't  ask about my projects at all, so that was a big L. As I left the interview room I knew my chances of getting selected where slim to none.
 

CVIT

CVIT interview went way better then I expected. After I went to the lab where CVIT interview was taking place the first guy had just come out.  .After 20-30 mins the second guy came out and my name was called. 

After I went they asked my name and told me to take my seat. There were again three profs and one of them was female she is most probably Jayanthi mam(P1)  The other two I think were Anoop sir(P3) and Narayanan Sir(P2) . Most of the questions were asked by P1 and P2 .P3 was mostly observing and provided inputs here and there.All three of them were very chill and funny.They asked for my SOP,resume. I had also brought my research paper with me. I gave them all three documents they went through it .They started with personal questions like where I am from , Where have I completed my UG from etc.Then P2 asked about my future plans.

me: After MS I want to work for a few years before going for a Phd

P2: Oh Phd very nice but why

me: Sir my dad has a Phd and that inspired me a lot and I decided long before( when i was in school ) that I will get my phd someday and I  am very much interested in AI research ,so i will definitely go for a phd in ML or similar domain.

P2 ,P1 was impressed that my father was a Doctrate. P2 then proceeded to ask about on which topic my father had completed his phd and what is he doing now. when they found out I studied in the same school my father is working they made fun of me and all three started laughing. P1 and P3 were going through my resume and SOP all the while .

P2: After Phd what will u do

me: it depends on the oppurtunities I have at that time sir

P2: will you not consider Teaching(joked about how carefully I shld answer this question as they were all phd's who chose to teach)

me: dont know sir (laughing) Teaching is definitely something I would want to try in the future . 

P2: Did u write GATE

me: yes sir mentioned my score (and then proceeded to tell them my gate flop story)

P1: Are You converted applicant or original

me: I converted mam

P1:you mentioned ur interest in research ,you also have a published paper then why dint u apply for MS initially ,wouldnt that be a better fit than Mtech considering your interests

me: answered how getting low gate score changed .my desicion.

They nodded understandingly and  talked about how less aware and hesitant students are when it comes to MS.I told them its not lack of awareness but uncertainity reg course duration,placements etc about  MS that makes students hesitant to apply and many students join Mtech for better placements . they joked on how if placements were their goal they should not join Mtech. everyone laughed. 

P1: what about you dint u get any oncampus offers( i had already mentioned i am dropper)

me: I had one from TCS mam , I kept it as my backup and I didn't join when they asked me to so they revoked my offer .

P2 again joked on how i am now in a situation i feared i will be in and living my worst nightmare(no backup options). Then they moved to my projects. For next  20 mins I explained my project (it is an emotion detection model that uses videos of actors to detect their emotions.CRNN's were used for feature extraction and a ANN for the detection) .both P1 and P2 asked questions about preprocessing , why I chose CRNN instead of a RNN or CNN . They were impressed that I wrote the code for ANN from scratch without using any predefined function.They then asked if my project worked on real time data I told no. P1 asked whether had I tried  I answered yes but as predicted it did not work. She laughed . They asked about the other project I explained briefly , then they moved on to the technical questions

P1: since u mentioned using back propagation do u know what is it

me: It is a set of equations which we use iteratively  to achieve optimum weights that minimizes the cost function , Since we travel from the last layer to the first its called back propagation.

P1: how do u get those equations 

me:using gradient descent

P2: what is gradient descent

me: we partially differentiate the cost function and use those equations to solve

P1: use the board and explain what it means geometrically

me went to the board drew a graph randomly and explained wht happens

P2: what do u mean by gradient show it using the graph

me: drew a slope for a point in graph and explained

P1: Now why do we call it descent

me:i was a bit clueless here I told since we reach minimum its called descent , All three of them gave hints and then i understood they were asking about the direction of the slopes and answered correctly.

P2: what will you do if we have to maximize

me: answered correctly

P2: ok what other subjects have you prepared 

me Linear algebra

P2: no tell me something else

P3:what was your favourite subjects during ur UG

me: I liked OS, ML,NLP sir

P2: oh OS we can ask in OS no problem for us

me:(laughing) no sir it will become problem for me I dint prepare OS sir.

P2: what about graph theory

me: I said ok (i dint want to keep saying NO)

P2: are u comfortable with shortest path algorithms

me: yes sir I know Djikstra's and bellman ford

P3: ok explain djikstra's

me: i used the board drew a weighted graph and started explaining the algorithm

P3 changed weights of some edges and  asked what will happen now etc.

after that all three seem satisfied and told me I can leave.

 Both my interviews got over by 12 so i  went back to the hall and volunteers told me I can leave .They had arranged lunch so i ate and left. After I went home I kept replaying the interview thinking where i could have improvised etc. but at the end of the day very little is in our hands. My only hope was CVIT and thankfully i got selected for CVIT itself. Some tips for future aspirants would be to be honest with the panel , dont lie they will figure it out in no time. CVIT panel focussed more on project even though I had used CRNN's their questions was on  ANN , They were genuinely interested in what i had done . You might revise the whole gate syllabus for the interview but the prof will find that one topic ur not strong in and ask you q on that just to see how u think and ans. dont get nervous ( main reason i messed up LTRC interview) be calm and confident. Finally, SOP,resume and luck plays an important role for getting shortlisted for interview as I know some ppl with high pgee scores not making it to the interview even tho they had publications.

ALL THE BEST!!

Thank you

    

290
290 views

Logistics & Arrival

Timeline: Reached IIITH at 7:00 AM. They didn't let us in until 8:00 AM. After verification, we were asked to wait in the Himalaya 105 block, where PhD and MS students informed us about our interview assignments. I was assigned to MLL and LTRC (my top two preferences). IIITH arranged breakfast and lunch for both parents and students.

Interview 1: LTRC Interview


two proffersors asked me question other two observed me

After i entered i gave my folder (with resume and sop) they asked me to sit 

 

P1: What is your mother tongue?

Me: My mother tongue is xyz.

P1: Let me give you famous newspapers of last one whole year, I would like you to get me the words in it.

Me: I will use a set and iterate through the paper, split the words, and add them to the set. (As a Python user, I would use a set)

P1: (Gives a confusing look)

Me: It's like hash sets as we don't keep duplicates. Or if I need a frequency, I would go for a dictionary.

P1: How does the hash set ensure no duplicates?

Me: I will take the hash of the words, check if it already exists, and if not, I will store it, sir.

P1: How do you compute the hash of the word? Compute the hash for a word in your mother tongue.

Me: We assign a numerical value for each letter like  a-1, b-2...  and multiply it with {#letters in lang+1)^position . Like "abc" would be $ 1*27*27 + 2*27 + 3  It might exhaust quickly as length increases, so it's better to use a long long data type.

P2: Why don't we use a float data type? It could hold a very large value, right? Find the value of this float value.

Me: (Calculates and provides the value) It could hold large values sir, but it can't store large consecutive numbers. It would start skipping values on large numbers as the mantissa is only 23 bits. (struggles to find the exact word)

P2: (After some time) It would lose its precision.

P1: Do you know what natural language?

Me: I don't know sir, but I have heard of it. Like the decoding of ancient scripts, or like in Project Hail Mary.

P1: Let's suppose you have to translate an alien language. How do you translate it?

Me: We have to know the alphabets, words, semantics...(i know i blabbered a bit) 

P1: You can leave now.

Panel 2: MLL Interview

3 panel members were there

asked me about where i come from i took a drop so i explained them why

they asked me what offers i have i told them IIITD mtech cse and BITS pilani Mtech cse

P1: So this is your first interview?

You: I had the IIIT Delhi research interview sir. (Starts shivering)

P1 & P2: You have offers already, so don't panic.

You: I would like to involve in research and do a PhD in future sir, so i am panicking.

these chats helped me relax a bit

P1: Which subject do you like?

Me: DSA, sir.

P1: What algorithms do you like in DSA?

Me: Heap and segment tree operations. Even though they are data structures, when I came to know about these, I wondered.

P1: Mention the algorithms that use a heap.

Me: Dijkstra, A* (at that moment i couldnt remember any other thing i sould have told prims and huffman coding) 

P1: Can you heap sort an array?

Me: I would heapify the array, then I will pop it out.

P1: Demonstrate it on the board.

Me: (Takes an array with arbitrary values and starts the heapify operation on the array itself) I would compare the element with the element at the n/2 position...

P1: Why are you comparing with n/2 element? Show us with a tree structure.

Me: (Converts the array to a tree structure and starts operating on it, but does a single pass instead of operating recursively, then directly removes the top element instead of swapping it to the last position)

Me: Realizing the mistake I should have done the heapify operation recursively. And instead of deleting the first element directly, I should have swapped it with the last element and bubbled down the top element by comparing its children so that I could have maintained the complete tree structure.

P1: Why are you comparing only with the children?

Me: By the heap property, the second minimum element will always be in level 2, sir, but that need not hold true for the third minimum element.

P2: What other subjects you prepared?

Me: DBMS and Linear Algebra, sir.

P2: How much on Linear Algebra?

Me: Till eigenvalues and vectors, sir.

P2 & P3: What's an Eigenvalue and Eigenvector?

Me: there are certain vectors(x)(should have said subspace of eigenspace) along the matrix(A) which it wont change the drection sir but it will expland or shirink by Evalue 

P2: (Asks follow-up questions)

Me: (Answers the follow-ups)

P2: Show us an example on the board.

Me: took a matrix [1,2][2,1] found its EValue as 3 (think the Evector is [1,1] - i dont remeber exactly now)

and showed them if you take them along the vector [x,x] => you will get [3x,3x] and took 1.[1,2]+1.[2,1]=[3,3] i have taken the x as [1,1] now the result is [3,3] 

P3: Can we solve it like that? (about this part 1.[1,2]+1.[2,1]=[3,3] )

Me: We could, sir.

P2: Do you know what a symmetric matrix is?

Me: (Doubtfully) A=Aᵗ 

P2: Do you know about its Eigenvalues?

Me: (Doubtfully) Real Eigenvalues.

P2: Prove it on the board.

as a follow up he asked me proove it on the board (started shivering) 

P2: helped me and asked me to do it with 2x2 matrix 

Me: took a 2x2 matrix  [a,b][c,d] 

P2: As it's symmetric, b = c right?

Me: changed [a,c][c,d] and try to find its eigen values (a-λ)(d-λ)-c^2 -> ad-(a+d)λ+λ²-c² -> λ²-(a+d)λ-(c²-ad) (started strugling) 

P2: How do you solve a quadratic equation?

Me: -b+root(b²-4ac)/2a and started using it(showing only part of it) (a+d)²-4(1)(-c²+ad) -> (a+d)²+4(c²+ad) started stuttering again 

P2: (Helps again by stating that the values are real to guide me to look closely at the equation)

Me: root((a+d)²+4(c²+ad)) as these are values under root and its a positive value it cant be a imaginary one -> so the solution should be real value 

P: You can leave now.

the symmetric matrix part is on youtube go classes LA lecture 9c - i learnt LA through youtube lectures

Post-Interview Timeline

The MLL interview ended around 1:00 PM. I had lunch, stayed on campus until 5:00 PM, ate the evening snacks provided by the college, and left around 5:30 PM.

i thought i have fucked it up in MLL and some what good in LTRC but luckily i got into MLL itself

 

Thank you

380
380 views

I had secured a GATE score 570 in 2025. My GATE 2026 preparation completely took a backseat due to severe time constraints, as I was working full-time as a Junior Research Fellow (JRF) at a DRDO lab. Therefore, I decided to apply for the Self-Finance M.Tech programme in the Department of Computer Science at NIT Warangal, using my valid 2025 GATE score.

The admission process started with a written exam on the 11th of June. The paper had 50 questions of 1 mark each, mostly covering Engineering Maths and other core GATE subjects. I managed to clear the written test, and around 300 students were shortlisted for the interview round in a 1:10 ratio. The interviews were scheduled over the next three days. My interview slot was on the last day, i.e., the 14th of June, in the afternoon session.

 

There were 9 panel members in the room, and all of them were subject experts. One professor was just taking notes, others were were actively asking questions.

First, they asked for my introduction and then asked about my favourite subjects. I told them: CN, DBMS, OS, C, and DS Algo.

P1: Do you know about Multiplexing and Demultiplexing? Me: I thought if I give a vague answer, they will grill me deeper, so I honestly said that I don't know.

P1: Okay, so what is the purpose of a Port? Me: I took a moment to think because I hadn't revised the subject properly. However, I have practical experience deploying software on offline servers in networks. Based on that, I explained that whenever you run a software or program on a server in a network, it needs to be assigned a specific port. This allows all computers in the network to navigate to it using the IP and Port combination. (P1 seemed satisfied)

P1: Then what is the function of an IP address? Me: I replied that IP is basically for logical addressing, used to uniquely identify a device in a network. Whenever a packet travels, the router checks the destination IP to route it.

P2: So tell me, what is the purpose of a MAC address then? Me: I told them that MAC is a 48-bit address integrated into the NIC (Network Interface Card). If we want to communicate with a PC on another network, we need an IP address. But to communicate with a PC within our own subnet, we use the MAC address. (P2 seemed satisfied)

P3: Do you know what Public IPs and Private IPs are? Me: I answered the basics but couldn't recall their exact significance, so I frankly told them that I am not able to recall.

P3: Explain the IP header and how fragmentation works. Me: I used the whiteboard to explain the IP header, specifically highlighting the key fields related to fragmentation like Flags and Fragment Offset.

P3: Let's say the transport layer has 4000 bytes of data and you have Ethernet which can transmit a maximum of 1500 bytes of data. How will fragmentation work, and what will be the fragment offset? Me: I drew the fragmentation process on the board and calculated the fragment offset for them.

P3: (Follow-up) Why did you divide by 8 while calculating the fragment offset? Me: I tried to convince them that the data must be divisible by 8 for scaling and rounding off purposes, but they did not seem satisfied with my explanation.

P2: Draw Congestion Control in TCP and explain its phases. Me: I drew the graph on the board, showing the slow start phase, congestion avoidance phase, etc.

P2: Till when will this slow start phase go on? Will it end anywhere? Me: Yes sir, it will end at the threshold. After reaching the threshold, it switches to the congestion avoidance phase.

P3: How is the threshold calculated initially? Me: I went completely blank for a moment. I told them we calculate it using Wc/2 (window size / 2) whenever three duplicate ACKs or a timeout happens. But they denied my justification and ended the interview by saying "Thank you!"

My Thoughts Post-Interview: I was quite unsatisfied with my performance in the last few questions and thought I wouldn't make the final list.

RESULT: SELECTED FOR CSIS! 🎉

497
497 views

Hello everyone,

I’ve spent the lot of time reading interview experiences on GateOverflow, always wishing that one day I’d be the one writing my own. Today, I'm incredibly grateful to share my journey, hoping it helps the next batch of aspirants handle the chaotic, exhausting, and unpredictable nature of the MS selection process.

My Profile

  • GATE Score: 686

  • Rank: 1366

  • Status: Final Year B.Tech Student

  • Shortlists Received: IITM, IITH, IITK, IITB, IIT Indore, and IIIT Hyderabad (via PGEE).

1. IIT Indore (MS Research) — The Wake-Up Call

  • Mode: Online (My first-ever interview)

  • Duration: 8–10 minutes (5 Professors)

I had read past GO experiences and thought IIT Indore would be a breeze because the questions seemed straightforward. I was wrong.

Being my first interview, I was incredibly nervous, and my responses weren't polished. I chose Linear Algebra (LA) as my subject, having studied Gilbert Strang post-GATE. I answered the first two questions regarding the span of vectors smoothly. Then came a question related to PCA. I hadn't formally studied it but tried to bluff based on something I had read briefly on ChatGPT.

The professor caught me instantly: "You didn't really study this, did you? You're just saying this by reading it somewhere." I was grilled for 2 minutes on my lack of depth. Later, a female professor asked a technical question about Gradient Descent, where I gave a generic answer instead of the mathematical depth she expected.

Lesson Learned: Never bluff in an IIT interview. If you don't know a niche crossover topic, just say so. I left this interview feeling completely defeated, convinced I had zero chance.

2. IIT Madras (MS) — The Numbers Game

  • Process: Written Test on Campus (June 6th)

The campus orientation was great, explaining the structure of the MS program. We were then taken to the lab for a written test.

The test itself was GATE-level (easy to medium difficulty), but the competition was brutal. Past GO posts often advise attempting around 20 out of 30 questions with high accuracy. However, due to the sheer volume of high-scoring candidates this year, the cutoff skyrocketed. I attempted 24 questions, but it wasn't enough to get shortlisted for the interview round. (Note: I had to skip IITH on June 7th because the timeline clashed heavily with my upcoming Mumbai dates).

3. IIT Kanpur (MS) — Coding & Concepts

  • Process: Morning Coding Round + Afternoon Written Test (No Interview this year)

I flew into Kanpur on June 8th.

  • Coding Round: Allowed in C or Python. I chose Python. There were 2 questions: one pattern-based problem and one Dynamic Programming problem where the brute-force approach surprisingly passed all test cases. Both were highly solvable.

  • Written Test: Medium to difficult GATE-level questions. It had 3 mandatory sections (15 questions each): Core CS, Theory, and AI/ML. You had to attempt everything.

4. IIT Bombay (MS & MTech RAP) — The Masterclass in Resilience

I arrived in Mumbai on the morning of June 11th, completely exhausted from non-stop traveling. Honestly, I only went to IITB to see the beautiful campus and had already booked my return flight for the 13th.

MS Written Test & Shortlist

The written test was genuinely hard, and given my fatigue, I performed just "okay." To my absolute shock, when results dropped that night, I was shortlisted! The CS panel alone had around 133 people, with another 100 split between IS and Theory.

The MS Interview Twist

On June 12th, I walked into the interview room with the HOD Ma'am. In my introduction, I mentioned my interest in AI. She stopped me immediately: "Why are you sitting in the CS panel then? We won't be doing AI work here." They were literally about to ship me out of the room.

Thankfully, I steered the conversation to my final year project—a choice my past self made specifically with MS interviews in mind. It was a Compiler Design project: a static analyzer built to analyze software design patterns via AST parsing.

Hearing this, the HOD paused and transferred me to a different panel with a professor who was an expert in that exact domain. When I met him an hour later, the interview completely changed. It felt like a passionate conversation between two engineers rather than an interrogation. I explained my logic, he asked about my thought process, and he even recommended a book to further my knowledge. I left the room deeply satisfied, happy that I hadn't embarrassed myself like with IIT Indore.

MTech RAP Interview (Web Dev Panel)

The next day, there was a coding round for RAP (C++ only). 4 questions: 2 easy, 1 medium, and 1 incredibly hard (requiring us to implement functionality into an existing system after reading an 8-page documentation manual). I solved 3 and gave up on the 4th with 20 minutes left.

In the subsequent interview, I was put in a Web Development panel. They asked about React. I hadn't touched React in two years and admitted my knowledge was rusty. When asked to write a simple code for a "Counter" application, the rust showed—I completely messed up the syntax of the arrow functions, and the interview ended right there.

The Ultimate Surprise (COAP Rounds)

After returning home, I was 200% certain I would be heading to a Top 3 NIT and applied for the CCMT counseling.

Then, COAP Round 4 arrived.

To my absolute amazement, I received two offers along with some 2nd gen IIT's:

  1. IIT Indore — MS Research

  2. IIT Bombay — MS (TA)

  3. IIT Kanpur - MSR (I received this offer almost a month after I accepted IIT Bombay through mail)

My Final Advice to Aspirants

  1. Your Project Matters: If you are applying for an MS, choose a solid, foundational project (like Compilers, OS, DBMS, or Core Systems) and know it inside out. It literally saved my life at IITB.

  2. Be Honest: If you don't know a topic, don't try to wing it with surface-level definitions you read on the fly. IIT professors will dissect your logic in seconds.

  3. Don't Count Yourself Out: Interviews can feel awful while they are happening, but professors are often testing how you think, not just what you memorize. Keep pushing through the exhaustion.

See you at IIT Bombay!

735
735 views
I did my BTech in Civil Engineering in 2025 and wrote GATE CS 2026. I decided to opt for MS programs in top IITs but unfortunately did not get call backs.

Luckily I wrote PGEE exam and did reasonably well I guess (They dont reveal the marks for MS students) so I was invited for MS interviews.

At IIITH the approach is different from that of other institutes. It is comprised of multiple research centres instead of a single CS department and these centres take the interview. I got invited to interview for my top 2 centre preferences, CSTAR (Centre for security, theory and algorithmis research) and MLL (Machine Learning lab)

1st Interview was MLL.

Panel:
P1:
P2:
P3:

The main lead was P1

Me: Entered. They did not tell to sit. Told me to erase blackboard (:P)
P2: (He was reading my details) When did you graduate?
Me: 2025
P2: (Smiling and looking at others) Is there a problem in the form? It shows his graduation date as 2003
Me: Sir, must be a glitch (Smiling sheepishly)
P2: I believe you, you dont look like you are 46 years old. (Corrects the date on the application)
P2: Where did you do your bachelors?
Me: I did my BTech in Civil Engineering from XYZ college. (P2's and other profs eyebrows raised when they heard civil engineering)
P2: So how much did you score in PGEE?
Me: I am not aware sir, the marks are not revealed to me.
P2: (Looks down, the marks were listed) You have got a very good score considering your background. Did you take any courses in CS in college?
Me: Yes sir, also I prepared for GATE for 6 months. Told all the courses i took.
P2: Interesting, what was your gate score and rank?
Me: 644 and 209X
P2: Thats a very nice rank. (I smiled and said thank you sir)

Meanwhile P1 was reading my SOP and resume to find the best area to roast me on ;)

P1: I see you have written you like graph theory in SOP. Can you tell me what is a matching?
Me: (stuttering)... Sir informally..
P1: (Smiles) this is a formal interview
P1: Can you write it on the board?
Me: I write the definition on the board, and they seem satisfied.
P1: Define a maximum matching
Me: Defined correctly.
P1: If we have a connected graph, what is the lower bound and upper bound on Maximum matching?
Me: (I take some time to think) sir in the case of lower bound (I said lower bound by mistake) each edge in the matching will match one pair of vertices, so maximum we can have n/2 such edges.
P1: isnt that upper bound
Me: yes sir, sorry i meant upper bound.
P1: ok now tell me the lower bound.
Me: (I start drawing some examples as I was not sure of the answer. I drew a cycle first) Told the maximum matching in a cycle which is a special case of connected graph
P1: that is correct but this is special case... tell me the general case
Me: (I was still not sure, this time I drew a complete graph, then I figured out that complete graph actually is the upper bound example) Sir in the complete graph we get the upper bound
P1: yes I know that (Little annoyed) Tell me how to get the lower bound
Me: (Now I started drawing a path graph... then I saw the connection...) Sir to this path graph if I keep adding edges to a vertex the matching size does not increase
P1: you are close.. complete the argument
Me: (Finally figured it out). Yes sir if I just add edges to a single vertex, the maximum matching will always remain 1 so lower bound is 1.
P1: Correct, which graph is this?
Me: Stuttering (Lol I cant believe I forgot the name of the graph)
P1: (Smiling) Isnt this the star graph?
Me: yes sir yes sir (Smiles awkwardly)
P1: ok now define an independent set
Me: Defined correctly.
P1: Draw an example and label the independent set.
Me: Drew correctly.
P1: Do you know what is a clique?
Me: Defined correctly.
P1: Tell me the connection between an independent set and a clique
Me: I said I am not sure.
P1: Do you know complement of graph
Me: Yes sir
P1: That is your hint.
Me: (Flustered again) I start drawing some more examples to buy time...
Me: Complete graph is also an example of a clique... if I take complement of a complete graph I get the empty graph which naturally is an independent set.
P1: Close, finish the argument
Me: (Realising) An independent set in a graph G is a clique in the complement G'.

Then the panel went to some math question which was very weird and cant really be explained here...

P3 finally speaks:

P3: Do you know algorithms?
Me: yes sir, I studied in GATE
P3: No I mean, have you implemented any classical algorithms like sorting, searching in code? Like in college?
Me: No sir, I have mostly done coding for software projects (Civil background)
P3: (Confused)... ok you can go now
P2: (Addressing P3) I think we can ask him a programming question
P3 (Thinks for a while): Ok write a code to reverse string
Me: Writes code correctly, P3 looks satisfied.
P3: Ok now you can leave. Take your documents.

This was a very weird experience because they grilled me on a concept in graph theory which even CS students dont study thoroughly (let alone civil students), and this was an ML lab so I expected ML questions or linear algebra (Had prepared more for that). I am not sure of selection.
Also they did not ask me anything about my resume, projects. They also seemed very strict.

Next interview was of CSTAR

Panel:
P1:
P2:

This time the interview felt more comfortable. They asked me to sit.

P1: Introduce yourself
Me: introduced.. again got same reaction when I said civil engineering.
P1: What subjects you are comfortable with. (Atleast he asked)
Me: Sir, I like linear algebra, algorithms...
P1: Ok you can go to the whiteboard.. he tells me to draw a 2*2 matrix.
Me: Drew
P1: Now lets say this matrix is M. How to find M^10?
P2: jokes.. why are you telling him to find M^10, let him find M^100 or M^1000
Me: (Sweating) Sir.. I think we can find eigenvalues of this matrix.. (I calculate eigenvalues)
P1: Ok...
Me: Then we know that if the eigenvalues of M is a,b then eigenvalues of M^10 are a^10, b^10
P1: That is correct. How to use that to find M^10?
Me: (I was stuck here as I did not know diagonalisation) I drew a general 2*2 matrix with 4 unknowns... then I wrote out the characteristic equation and plugged each eigenvalue which was calculated from M
P1: Interesting approach... but wont you get only 2 equations? How will you solve for 4 unknowns? (I think he was testing me because he had purposely given a symmetric matrix which I did not notice. In that there are only 2 unknowns, not 4)
Me: (I was stumped here)...
P2: Do you know diagonalisation?
Me: no sir.
P2: Ok... then leave this approach.
P1: Still, its a nice approach, try doing it later
P2: Tell me if you just had to quickly solve M^10 in an exam what will you do?
Me: Sir I can repeatedly square it.. wrote M^10 = (M^2)^2^2 * M^2
P2: So how many operations?
Me: 4
P2: Ok tell me another subject
Me: Sir, graph theory and algorithms
P2: Define dijsktra algorithm
Me: Told
P2: Does it run with negative weight edges.
Me: Sir, while the algorithm is not intended for negative weight edges, there are some special cases where negative weight edges will still yield a correct answer.
P2: interesting. can you draw an example
Me: (Struggling to draw)
P2: (Gives a hint, kind of). Draw any graph which has -ve edges.. show me dijkstra on it
Me: Draws and shows. then realises that graph does not terminate if the -ve edge is on the shortest path
me: Sir algorithm will not terminate as the negative edge is on the path.
P2: yes, (Then he explains a full story about how shortest walks itself are ill defined and that is why we cannot solve for negative edges using Dijkstra)
P2: Now what about MST. Define MST.
Me: Defines.
P2: Can we run any MST algorithm on -ve edges?
Me: No sir. (This is wrong answer but for some reason he continued and did not correct me)
P2: Ok now let me define a maximum spanning tree as the spanning tree with maximum cost of edges. I want to find the maximum spanning tree of the graph but I cannot write my own algorithm, I can only use the traditional MST algorithm like Prim/kruskal. The only thing I can do is modify the weights of edges.
P2: (Continues). Naturally one would think to invert the edges and make them negative but you just said MST algorithm cannot run on -ve edges. So what to do?
Me: (Stumped) Draws some example to buy time.
P2: Think of K's complement
Me: it is some w - number.
P2: Ok so how do we invert the edges but keep them positive so the MST algorithm can work
P2: (He does not wait for my answer) we can just take max of all edges and subtract the weight from it right?
Me: I nod ;)
P2: ok good. Tell me one more subject
Me: Sir, I am good with algorithms
P2: Ok do you know recurrences?
Me: i can try my best to solve sir
P2: Ok write a(n) = a(n-1) + a(n-2) does this look familiar
Me: Yes, it is fibonacci recurrence
P2: Ok solve and find closed form.
Me: Solved.
P2: Ok come and sit

Then the panel grills me on my personal motivations for more than 15 mins which I wont talk about here.

Each interview felt like almost an hour. I dont know if it is because I I am dumb but the questions were quite difficult to me and required me to think a lot, most other people did not have the same experience as their panels focused on the projects and resume more. (I have a data science internship so i was hoping for questions on that). I also did not expect graph theory in an ML lab interview (,,>﹏<,,)

Results are not out yet. I am not very hopeful of selection tbvh.

Edit: I ended up getting selected for CSTAR :))
247
247 views
Total seat = Around 10

Interview taken by Prof Himanshu Yadav

Q1.Introduce yourself and tell your research interest

Q2.What is an Artificial Neural Network?

Q3.How does Backprogragation and Gradient Descent works?

Q4.Some question on Loss function, trigonometry and probability

Q5.How to compare two computional models?

Q6.What is a P600 waveform?

Q7.Explain Cross Validation in Machine Learning?

Q8.Why there is a need to split dataset into training and testing set? What's the main reason?

 

Advice given by sir to me :- In the field of AI/ML having theoretical knowledge is good but practical knowledge is more needed.
1,476
1,476 views

IIT Madras DSAI MS (Research) Interview Experience — Detailed Writeup

"PS: I have used chatgpt to strucuture it properly it is not AI Slop"

I’m sharing my complete experience of the MS (Research) admission process at IIT Madras DSAI (Wadhwani School of Data Science and AI). I hope this helps future aspirants understand:

  • the written test pattern

  • interview expectations

  • preparation strategy

  • and most importantly, what the interviewers actually seem to value.


Background

I prepared for GATE CS for roughly 2 months:

GATE Performance

  • Score: ~555

  • Marks: ~48

  • Rank: ~4000

  • Category: General

Honestly, I was quite disappointed because in the final few GO Classes mocks/AIMTs I was consistently scoring 60+, but that performance did not translate into the actual exam.

Still, I had heard that interdisciplinary AI programs such as: IIT Madras DSAI, IIT Delhi Yardi School sometimes have comparatively lower cutoffs and more holistic selection criteria, so I applied.


Application & Shortlisting

I received written test/interview calls from:

IIT Madras DSAI (MS Research)

  • Online written test

IIT Delhi Yardi School (MS Research)

  • Offline test at physical centers

I did not get shortlisted by IIT Delhi later, which I believe may be because they give relatively higher weightage to GATE score (same observation by other people too)


IIT Madras DSAI Written Test

After GATE, I wasted almost 2 months (February and March). From April onward, I started revising:

  • Linear Algebra

  • Probability

  • Basic Machine Learning

Initially this preparation was for IIITH PGEE, but it helped significantly for IITM DSAI as well.

Test Statistics

  • Around 800+ students appeared

  • Approximately 90–100 students were shortlisted for interviews

Test Pattern

The paper had approximately:

  • 35 MCQs ( Linear Algebra, Probability, Calculus, ML/Optimization, Programming)

  • 3 coding questions (to be solved in Python and were relatively manageable)

I attempted:

  • ~32 MCQs

  • Solved all 3 coding questions comfortably


Important Observation

If:

  • your GATE score is around 550

  • and you are genuinely interested in AI/ML research

then definitely apply to:

  • IIT Madras DSAI  &  IIT Delhi Yardi School

Even with an average GATE rank, you still have a realistic chance of getting shortlisted.


IIT Madras DSAI Interview Experience

The interview was technical, but not designed to eliminate students through obscure puzzles. The panel mainly focused on:

  • mathematical intuition

  • conceptual clarity

  • geometric understanding

  • ability to reason under pressure

  • honesty about what you know and don’t know

The interview lasted around 20-25 minutes and my panel had 3 professors.


Initial Discussion

The interview began with:

  • self introduction

  • why I left a 30+ LPA role and why I wanted to pursue MS Research

The discussion was fairly conversational initially.


Professor 1 — Linear Algebra & Geometric Intuition

This was by far the deepest and most interesting part of the interview.

The professor focused heavily on:

  • geometric intuition

  • vector spaces

  • transformations

  • understanding beyond textbook definitions

Questions Asked

1. Explain Row Space and Column Space

I went blank and forgot the textbook definations (that they are space spanned by LI rows/cols), so I explained whatever logic was comming to my mind I explained  in terms of geometric interpretation(that row space space consisting of all vectors which gets transformed to a non zeero vector in column space during Linear Transformation).

2. What is Null Space?

I again went off the real defination, I somehow forgot everthing i knew, I intutively explained that it cointains all those input vectors which can't be mapped to column space, thus it is all vectors in space formed by dimensions which got destroyed/squished in Linear Transformation)


3. How are Row Space, Column Space and Null Space related?

The professor was clearly trying to test geometric understanding rather than memorized definitions. I copy pasted whatever i had learnt from gilbert strang's video on 4 fundamental spaces few days ago.


4. Can a matrix have common Null Space and Column Space?

I thought:

  • null space belongs to the domain

  • column space belongs to the codomain

I  concluded they could never be same space physically.


5. Under what condition can Null Space and Column Space are similar? Say for a 2x2 matrix what will be the constraint?

I eventually realized:

  • rank = 1

  • nullity = 1

would be necessary.


 

5. b: [edit]: He asked the followup on Rank Nullity theorem and What do I understand by it?


 (I  explained it geometrically in terms of information preserved vs information destroyed by Linear Transformation, Dont remember exact Discussion here)


6. Construct a 2×2 matrix whose Null Space and Column Space coincide

This became an extended discussion.

Initially I proposed an incorrect matrix. The professor asked me to identify why it was wrong. I realized:

  • I had accidentally made the spaces orthogonal instead of identical.

This was actually a very important learning moment.


7. Are Column Space and Null Space always orthogonal?

In panic, I incorrectly said yes. {At this point i had not given a single precise answer to any question in any past questions (Atleast what i felt), I was too afraid to think, I said yes ,as in back of my mind i remembered Null space perp to row space, which i had explained earlier)}


8.  He corrected me "what if i say they are not"?

I reslised my mistake and explained why it is not, " yes they are not always perpendicular,  the null space is always perp to row space"

9. He asked "Are you sure Null space and Row space are perpendicular"?

I thought and was able to come up with this "rank is no of dimensions which matter in transformation (which is also the dim of row space, by mattering we means every vector of roe space is being transformed to a vector in column space)  so it is wrt row space and nullity is dimension which get destroyed so they both represent different dimensions of domain space, and since two dimensions in R^n are always perpendicular so thats why null space and row space are perpendicular"


10. So now you have this case where col space and row space are Orthogonal, and you need them to overlap how can you modify one of them keeping other same? (He said this is his last question and gave me 1 min to think)

After some thought, I was finally able to:

  • construct the correct matrix and once it clicked,  I was able derive the general condition for such 2×2 matrices.


My Biggest Realization from This Section

At this point, I was completely convinced I had ruined the interview.

I had:

  • forgotten textbook definitions (like who the hell forgets the defination of row and col space, it's the first thing we learn)

  • made multiple mistakes with things, definations i should know by heart

  • needed hints several times

I even felt emotionally devastated after the interview.

But looking back, this section probably helped me the most.

Why?

Because the panel was not looking for:

  • perfectly memorized definitions

  • polished textbook answers

They were evaluating research ability:

  • whether I could think mathematically, and right from start I always gave answer as if I was asnwering based on the concept and Geometric interpretation , they were not the precise definations and might have left few cases here and there, but it was demonstrating reasoning,  formation of hypothesis from what i know geomentrically etc.

  • whether I had geometric intuition

  • whether I could reason after receiving hints

  • whether I could recover from mistakes

Even though my answers were imperfect, I kept trying to reason visually and conceptually.

That mattered much more than I realized during the interview.


Professor 2 — Machine Learning

This section was comparatively shorter.

Questions Asked

1. Tell me about CNNs? 

(Me: I have not studied them properly)


3. Explain:

  • supervised learning

  • unsupervised learning


4. Explain K-Means

I only knew the high-level idea explained it and before cross questioning would start I admitted that openly so prof didn't went to far here, just asked why it is unsupervised.

5. Have you done any ML courses?

I honestly answered:

No formal course, but I had recently started studying ML seriously. The professor was slightly dissatisfied, probably because:I had a Tier-1 CSE background, but limited formal ML exposure, he even asked what was my major 😭?

Still, honesty was better than pretending.


Professor 3 — Algorithms & Problem Solving

This went significantly better for me.

Questions Asked

1. Difference between Merge Sort and Quick Sort


2. When would you prefer Merge Sort over Quick Sort?

I discussed:

  • external sorting

  • sorting files larger than RAM

  • multi-way merge concepts

This led to a short discussion on m-way merging.


3. Two Sum Problem

I explained:

  • brute force approach

  • hashing approach

  • sorting + two pointers

  • complexity comparisons

The discussion here felt much smoother and more natural.


Final Result

SELECTED


What I Think IITM DSAI Actually Values

After reflecting on the interview, I think the panel cared far more about:

1. Mathematical Thinking

Can you reason through unfamiliar situations?


2. Geometric Intuition

Especially in Linear Algebra.

Not just:

“column space is span of columns”

but:

  • what transformations are doing geometrically

  • how spaces interact

  • what rank/nullity actually mean visually


3. Recovery After Mistakes

They deliberately pushed discussions deeper after mistakes.

They seemed interested in:

  • whether you freeze

  • or whether you continue reasoning.


4. Honesty

Whenever I didn’t know something in ML, I admitted it directly.

That is much better than bluffing.


5. Problem Solving Ability

Especially in algorithms and mathematical reasoning.


Topics I Strongly Recommend Preparing

Probability

  • Random variables

  • Bernoulli/Binomial/Geometric distributions

  • Expectation

  • Conditional probability


Linear Algebra

  • Vector spaces

  • Basis & span

  • Rank-nullity

  • Linear transformations

  • Row/column/null spaces

  • Orthogonality

  • Determinants

  • Eigenvalues & eigenvectors

Most importantly:
focus on geometric understanding.

Gilbert Strang’s lectures are extremely useful for this.


Machine Learning Basics

At minimum:

  • Linear Regression

  • Logistic Regression

  • Gradient Descent

  • Bias-Variance Tradeoff

  • Bagging

  • K-Means

  • MLE basics


Final Advice

Research interviews are often exploratory discussions.

The professors are trying to understand:

  • how you think

  • how you reason

  • how you respond to guidance

not whether you can recite perfect textbook definitions under pressure.


 

Questions I Collected from groups(asked to other peeps) or From Online sources for my prep:

 

Basic Probability

1. What is a random variable?

Expected idea:
A random variable maps outcomes of a random experiment to numerical values.

Example:
In a coin toss experiment, define:

  • X = number of heads

This converts outcomes into numbers that can be analyzed mathematically.


2. Difference between a random experiment and a random variable


3. Difference between a random variable and an event


4. What does “fair” mean in a die roll?


5. Difference between “fair” and “equally likely”


6. Expected number of throws until a 3 appears

Follow-up:

  • Derive it mathematically step-by-step

7. Difference between Bernoulli and Binomial distribution


8. Probability of getting at least one head in 5 coin tosses


9. What is a geometric distribution?


10. Mean/expectation of geometric distribution


11. Expected number of tosses required to get two consecutive heads


12. What is π?


13. How can π be estimated using Monte Carlo methods?

(Hint given: square with an inscribed circle)


14. Language/alphabet probability question

A language has 5 alphabets with probabilities:

  • p1, p2, p3, p4, p5

What is the expected number of times alphabet 1 appears as the first letter?


Machine Learning Questions

1. Which ML topic are you comfortable with?


2. What are the steps involved in simple linear regression?


3. If data has 5 dimensions, what do we fit geometrically?

(Expected idea: hyperplane)


4. What is linear data?


5. What is gradient descent?


6. Why is the learning rate α used in gradient descent?


7. Difference between bias and variance


8. Explain underfitting and overfitting


9. What is the bias-variance tradeoff?


10. What is bagging and what problem does it solve?


11. Basics of:

  • Linear Regression
  • Logistic Regression
  • Softmax Regression
  • Maximum Likelihood Estimation (MLE) {2-3 adviced me this was of utmost importance, may be if you go into Cost funct discussion they will ask this}

Linear Algebra Questions

Vector Spaces & Basis

1. What is a basis?


2. What is the span of vectors?


3. What is ( R^n )?


Linear Transformations

4. What is a linear transformation?


5. Why is it called “linear”?

(Expected intuition:

  • lines remain lines
  • origin remains fixed)

6. When does AB = BA hold?


Determinants

7. Geometric interpretation of determinant


8. Why is:

det(AB) = det(A)det(B


Gaussian Elimination

9. What spaces remain unchanged under Gaussian elimination?


10. Why should Gaussian elimination not be used directly to obtain a basis for column space?


Linear Independence & Rank

11. What is linear independence?


12. Relation between null space and linear independence


13. Conditions for:

  • No solution
  • Unique solution
  • Infinite solutions

(using rank arguments)


14. Why does determinant 0 imply infinitely many solutions may exist?


Orthogonality & Subspaces

17. What is orthogonality?


18. Why is row space orthogonal to null space?


Eigenvalues & Spectral Theorem

19. Applications of eigenvalues and eigenvectors

Examples:

  • Fibonacci sequence

20. State the spectral theorem


HR / Motivation Questions

1. Why do you want to pursue MS instead of a job?


2. If selected, what area would you like to work on?

Resources I followed:

I would recommend to go and watch first 3 lectures of CS229-2019(Anand Avati one) He has mentioned all the Important concept of LA and Prob which we need to understand, Just make note of what topics/concept he mentions, then use any good resource to get deeper grasp on them. 
I used these resources:

For LA:

1. Gilbert Strang Lec (MIT 18.06(not all lectures just important concepts) and 18.065 (LA for ML concepts understanding))

2. 3blue2brown is Must watch

Note: I watched Sachin sir's lecs too but that was for pgee, 18.06 is more then enough

For PROB:

1. I didn't studied any specific course, just watched the videos (or chatgpt) on topics mentioned by Anand Avati on youtube (mostly MIT 6.1200J + statquest ocassionally)

FOR ML:

1. CS229 is good, I personally watched mix of 2018 (Andrew NG) and 2019(Anand Avati), Anand explains more statistically ( mathematically), And well Andrew is Andrew.

 

If your fundamentals are decent and you can think honestly and mathematically, you still have a strong chance.

Hope this helps future aspirants.

All the best!

587
587 views

Hello Everyone, I will keep it short and sweet.  

I am Deva, I wrote gate cse in 2025, Gate rank 486, Score 772

For mtech ra in iitb there is a compulsory programming test, we had 4 questions in total, i could answer 2 of them, after which i was shortlisted for interview, the questions will be there on this link.

some tips for programming test: we will be using iitb's inhouse teaching website to give the test(bodhi i think it is called), we will work on ubuntu system, gdb will be there, vscode will be there(no extensions), if you know how to configure gdb for vscode good, otherwise you can do debugging in cli anyway. 

you will be given reference pages for basic i/o operations, memory operations those kinds of things, NOTE that c++ STL reference will not be given, you cannot use cppref or other sites, you need to know them by heart. 

Here is what i remember from my interview - i have excerpted only relevant parts from what i remember - and one quick note - for sysad interview final year students will only be interviewing you since they only have to do the KT to you, so i guess institute lets them take the interview, there was one prof in the panel (i think it was soumen chakraborty sir) , and the vibes were very chill.

q1: introduce yourself 

 i had prepared a strong introduction, since i was expecting this question 

told about my project, and my work ex. in that the last point was as follows 

...My main contribution was in developing a memory module for a text-to-sql chatbot for a logistics company

 

q: can you tell us about this memory module, as much as you are permitted to reveal ofc

I told them about it

 

q: in your schema, was the schema tailored to the client or was it generalized

a: initally we had planned of making it generalized but later on because of time constraint we ended up making it specialized

 

q: give an eg 

( an eg)
 

q: how much ai did you use in your coding?

a: i have switched to doing unit testing completely using ai, it saves a lot of time, after the tests are ready i debug them to ensure they are actually testing what they are supposed to test. 

and in general it completely depends on time constraint if we need to ship something today then full thing we do on ai, but i try to do thing by myself as much as possible. 

(then i asked the prof )

 

q: sir what do you think, how do we balance using ai and coding manually? what are your thoughts on that subject.

I told them about it

a:  he said something about how for newer languages and newer uses ai is very bad, he said some specific egs also i forgot
 

q. which language do you like?

a: most of my work i have done in python but i prefer c++

 

q. what do you think are cons of python?

a: speed obv

(then he started talking about programming languages, he was saying that origin of python was like they had a plan and make python according to that. normally how new programming languages are made is , there is a problem and then we try to solve the problem using  the new prog lang, and that python did not evolve naturally like that)

q: which filesystems have you used?

a: mainly ext4, but i have read about zfs, and one of my seniors was working on btrfs so i sat with it for a while

 

q: what is the usecase of zfs?

a: .....snapshotting .....

 

q: if we were to try to implement snapshotting in ext4 how would we go about doing it?

a:  ...knew 1 simple way told about it....

 

q: then he asked some active question in zfs that i dint even understand, something like -  there are block devices in linux, zfs can pool all these devices and make it one block device, and then implement RAID, these are 2 diff things could they have not been seperated, why did they go with this design choice basically, what are you thoughts on this? 

a: ...some gas, i think he was fine with it......


q. can you tell us about how the packet travels through the network stack when receving?

told about napi, hard req, soft irq, and path in general

 

q. what ways did you look at in improving your latency?

talked about rss, rps, rfs, arfs 

then i got prompted about kernel params like buffer sizes then i mentioned that i used that also.

 
 
thats it then they told me you can leave.
626
626 views
Myself Maitrey Patel , I have given Gate CSE 2026

GATE Score - 743

GATE Rank -707

IIT Madras called all the shortlisted candidates for the MS CSE interview process on 6th May. We had to report at 8:00 AM at the CSE Department. Indian Institute of Technology Madras Professor Madhu Mutyam gave an introductory speech regarding the MS and PhD programs. After the speech, we were taken to the labs for the written examination.

This time, there were two written tests: W1 and W2, out of which we had to choose one. Students selecting W2 were eligible only for the Theory Pool interview if shortlisted. Students selecting W1 could choose any two out of the three available pools for interviews if selected in the written test.

The 3  Pools are (A) Theory (B) Systems (c) Intelligent system

MS W1 topics: The exam will be based on core areas of Computer Science and Engineering, including C programming, digital logic, discrete mathematics, data structures, algorithms, computer organization/architecture, theory of computation, operating systems, database management systems, compiler design, and computer networks. In addition, there will be questions on analytical ability.

MS W2 topics: Theory of Computation, Data Structures, Design and Analysis of Algorithms, Combinatorics, Logic, Graph Theory, Linear Algebra, Probability, Analytical ability.

The written test duration was 2 hour and contained 30 MCQs. The exam finished around 12:00 PM, and the results were announced around 2:00 PM. My name was on the shortlisted list. The interviews were scheduled from the next day, i.e., 7th May. Some students had their interviews on the evening of 6th May due to overlapping interview schedules with other IITs.

The interviews began at 9:00 AM on 7th May. I had selected the Intelligent Systems (AI/ML) pool and the Systems pool for interviews. My Systems interview was scheduled at 2:00 PM on 7th May, and the Intelligent Systems interview was scheduled at 9:00 AM on 8th May.

Before the interview, we had to complete document verification. The department staff verified documents such as 10th and 12th marksheets, all semester marksheets, GATE scorecard, category/EWS certificate, and college degree certificate (if available).

We were sitting in Aryabhatta Hall when a department staff member called my name and took me to the meeting room where the interview was going to happen. As soon as I opened the door, Professor Christo Rebeiro and two other faculty members were sitting there. They asked me to sit down and relax.

Systems Interview

Professor 1: Select one subject from the list containing OS, COA, DBMS, CN, and maybe Compiler Design as well.

Me: OS

Professor 1: Explain Binary Search Tree.

I explained BST orally and also drew its structure on the board.

Professor 1: Now write a C program to perform search in a BST and also define the binary node structure in C.

I wrote the complete code for searching in a BST using a single search function and also defined the BST node structure. They did not ask me to write #include statements or the main() function.

I had to write the code on a smartboard using a pen, so they were not concerned about minor syntax errors.

Professor 1: Explain your code and tell the time complexity.

I explained the code and said the time complexity would be O(log n).

Professor 1: How do you know the time complexity is O(log n)?

I wrote the recurrence relation and explained that every time we reduce the search space by half.

Professor 2: What topics are you familiar with in Operating Systems?

Me: Synchronization, scheduling algorithms, paging, etc.

Professor 2: Let’s talk about process scheduling. What are SJF and Round Robin?

I explained both scheduling algorithms.

Professor 2: Is practical implementation of SJF possible?

Me: No sir.

Professor 2: Then why do we study it?

I did not have an answer, so I just smiled, and he smiled too.

Professor 2: What happens in Round Robin when the time quantum completes?

Me: Context switching occurs, and the next process in the queue gets scheduled.

Professor 2: Explain context switching in detail in a uniprocessor system. What exactly happens when the time quantum expires?

I explained the whole process of context switching. First, an interrupt occurs, then the values of registers, program counter, etc., are stored in the PCB. After that, the values of registers and program counter of the next process are restored from its PCB.

Professor 2: Who stores the values of registers and program counter into the PCB — the processor or the OS?

After thinking for a while, I answered:

Me: The OS, sir.

Professor 2: How does the OS store values in the PCB?

Me: The OS is also a program. It has instructions that execute and store the register values into the PCB.

Professor 2: So you are saying the processor is executing process P1, and suddenly, by magic, the OS instructions start executing on the processor?

At that moment, a staff member came in with tea and coffee for the professors. After that, the professor also asked me whether I wanted tea or coffee.

I answered that there would be a system call after which the OS instructions would execute on the CPU.

Professor 2: Now you mentioned the term “system call.” Tell me what a system call is.

I explained system calls with examples.

Professor 2: Now come back to the previous question and answer it again.

Me: I do not know the exact answer.

Professor 2: Okay, then just guess what might be happening.

The professor was laughing while asking this.

Me: I again answered, “System call.”

Professor 2: You can go now.

As I was standing up and about to leave:

Professor 2: Suddenly asked, “Why do you want to do an MS?”

Me: Sir, I want to solve real-world problems and make the lives of common people easier.

Professor 2: You may go now.

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Intelligent Systems (AI/ML) Interview

On the next day, 8th May at 9:00 AM, I had my Intelligent Systems interview.

I was again sitting in Aryabhatta Hall when a department staff member called my name and took me to the meeting room.

As soon as I opened the door, two professors were sitting there.

Professor 1: Good morning, Maitrey! How are you?

Me: I am good, sir. How about you?

Professor 1: I am also good. Have you had your breakfast?

Me: Yes, sir.

Professor 1: Be relaxed. We can start the interview now.

Professor 2: Select one subject from the list. The list had subjects like Linear Algebra, Probability, and Data Structures.

Me: Linear Algebra.

Professor 2: Take a marker and write a C program to check whether two given vectors are linearly independent or not.

Input: Two n-dimensional vectors.
Output: true if the vectors are linearly independent, otherwise false.

I started writing the code in C. Initially, I was thinking in terms of row echelon form and row operations.

Professor 1: Think in terms of the condition for two vectors to be linearly dependent.

After thinking for a moment:

Me: Sir, they should be multiples of each other.

Professor 1: Smiling, “Implement this.”

I implemented it using a loop where I divided the elements of one vector by the corresponding elements of the second vector. If the ratio changed at any point, I broke the loop and returned false.

Professor 2: Okay, I am done.

Professor 1: Let us start with Linear Algebra. Explain eigenvalues and eigenvectors.

I explained everything I knew about eigenvalues and eigenvectors.

Professor 1: How do we derive eigenvalues?

I explained the complete derivation process on the board.

Professor 1: Take a 3×3 matrix A whose all entries are 1. What will be its eigenvalues? Answer without calculation.

Me: Sir, 3, 0, 0 because the trace is 3, and there is only one linearly independent column.

Professor 1: What will be the eigenvector corresponding to eigenvalue 3? Answer without calculation.

I started thinking and mentally trying to calculate it.

Professor 1: Don’t calculate mentally. Think about the approach you mentioned two minutes earlier.

Me: Is it related to linear combinations?

Professor 1: Yes. First try for eigenvalue 0.

Me: Sir, one eigenvector can be [1, -1, 0].

Professor 1: What will be another eigenvector for eigenvalue 0?

Me: Sir, [1, 0, -1].

Professor 1: Now try for eigenvalue 3.

Me: Sir, [1, 1, 1].

Professor 1: Are you sure?

Me: Yes, sir.

Professor 1: Okay, your interview is over. You may go now.    ​

The IIT Madras administration also provided hostel accommodation. They were giving single rooms, but I requested a shared room with my friend, so they allotted us a shared room in Krishna Hostel. The total accommodation charge for 4 days was around ₹2700, which also included mess food.

The mess food was included in the charges and covered breakfast, lunch, and dinner. The breakfast at Nilgiri Mess was especially awesome. You could find deer and monkeys everywhere on the campus. The campus atmosphere was beautiful and gave an old-school vibe. The CSE department also offered us tea, coffee, and snacks multiple times during the interviews.

Result - Still Awaited

contact me - https://www.linkedin.com/in/maitrey-patel-571041249/
507
507 views

There are 2 steps in the admission, first being a written test, followed by a single 40-45 minute interview both in-person at the beautiful IITM Campus.
Direct-PhD applicants had their interview on the same day as the written test, or on morning of next day. MS applicant's interviews were held on the next day, unless somebody had other constraints. The cutoffs and more details available here.

Written Test Pattern: 30 MCQ questions in 2 hours with negative marking. The level of question were on par with GATE.
I appeared for the W2 test. The tests for MS and Direct-PhD applicants appeared for the same set of tests but separate shortlists.

MS W1 topics: (choose any 2 of 3 of the panels)
The exam will be based on core areas of Computer Science and Engineering, including C programming, digital logic, discrete mathematics, data structures, algorithms, computer organization/architecture, theory of computation, operating systems, database management systems, compiler design, and computer networks. In addition, there will be questions on analytical ability.

MS W2 topics: (only panel 1)
Theory of Computation, Data Structures, Design and Analysis of Algorithms, Combinatorics, Logic, Graph Theory, Linear Algebra, Probability, Analytical ability.

The 3 panels were: 
1) Theoretical CS
2) Computer Systems
3) AI & ML

More details about the interview click here.

Here goes the complete details about the interview.

Prof. Shweta Agrawal reads my educational details and reads my SOP.
Asks me a couple questions from there to make me feel comfortable.

Then she asks me to choose 2 topics from the list and for each topic there were professors who are specialized in their respective domains who would ask questions.
More than 10 professors were present in the room.

Topic 1 : Discrete maths Set - Theory and Counting (Prof. Raghavendra Rao BV)

1. What is Bijective function? 
2. If the 2 sets are the same and they are one one, are they also onto?
3. Can you Proof answer of 2?
4. What is countably inf and uncountably inf set. 
5. Definition of countably infinite.

Topic 2: Graph theory (Prof. Nishad Kothari)

1. What is Coloring in graph?
2. What is 2-colorable?
3. What is a Bipartite graph?
4. Are 2 and 3 same?
5. Then, how can we convince someone a graph is not bipartite?
6. Concept of non existence of odd cycles came here, to show not bipartite.
7. What is Tree?
8. What is a Leaf edge?
9. Does a tree always have a leaf? 
10. Apart from null graph does all tree have a leaf?
11. Can you proof statement in 10?

Additional Questions: (Prof. Aishwarya Thiruvengadam)
1. You worked on implementing threshold secret sharing, what was it? (mentioned in my SOP)
2. Follow up question about what else I did.

Then they asked me if I had any questions for the panel.

Hope this post has been helpful, please feel free to ask me questions or queries in the comments. 😄
But please don't ask me how my interview went or any other personal questions. 🙏🏻
(This post is also written without any AI help, so please excuse grammatical mistakes)

634
634 views

I appeared for the IIT Hyderabad Self Sponsored MTech CSE/NIS written test conducted on 8th May 2026.

The institute kept the question paper with themselves, I noted down as many questions as I could remember after the test. These are memory-based questions so bear with me if there are any mistakes in the wording of a question. 

Test Pattern

  • Duration: 1 hour 30 minutes
  • Total Questions: 30
  • Each question carried 1 mark
  • Negative marking: 0.25
  • Total Marks: 30

I was able to reconstruct around 24-25  questions

 

Questions

Q1. AVL Tree

What is the recurrence relation for the minimum number of nodes in an AVL tree of height h?

 

Q2. SCC and DFS Finish Time

A directed graph has three strongly connected components A,B and C

There are directed edges:

  • A→B
  • A→C
  • B→C

Question was related to the ordering of DFS finishing times among these SCCs.

 

Q3. Maximum Nodes in a Heap

What is the formula for the maximum number of nodes in a binary heap of height h?

Q4. DFS Discovery Time

Consider a graph with vertices numbered from 0 to 80.

Two vertices i and j are connected iff:

|i − j| ≤ 2

Adjacency lists are stored in decreasing order.

For example, adjacency list of node 1 is:

1 → 3, 2, 0

DFS starts from node 0.

How many vertices have discovery time greater than the discovery time of vertex 40?

Q5. MST Question

Given a connected weighted graph G(V,E) whose edge weights are not necessarily distinct, determine which statements are true.

I remember 2 options:

  • The minimum weight edge belongs to some MST.
  • The maximum weight belongs to no MST.
 

Q6. OS

Initial value of x = 0.

Operation x++ is executed twice in parallel and increment operation is not atomic.

How many possible final values of x are possible?

 

Q7. Cache Misses

In which of the following cache organizations do conflict misses occur?

  • Direct Mapped Cache
  • Set Associative Cache
  • Fully Associative Cache
  • Write Through Cache

Q9. Cache 

Question involving calculation of:

  • Tag bits
  • Index bits
  • Offset bits

for a 2-way set associative cache. Main memory size was 32MB, Cache size was 32KB and Block size was 2^6 Bytes.

Q10. SQL Query

3 tables were given. Movies(Mid,Name), Actors(Aid,Name) and CI(Mid,Aid). We had to find Names of all movies in which all actors in the database participated. 

 

Q11. C Language Question

Given the following structure definition:

 
struct A {
    int x : 3;
};

struct A a; 

Which operation cannot be performed on x?

Options included expressions similar to:

  • x++
  • &a.x
  • a.x = 5

Q12 & Q13 : 2 Compiler Design theoretical questions. 


Q14. Operator Precedence

Given:

int x = 5;
x << 1 + 1

Find the value of the expression.

Q15. Left Outer Join

Is LEFT OUTER JOIN:

  • associative?
  • commutative?

Q16. Functional Decomposition

One decomposition related DBMS question.

Four statements/options were given and we had to identify the false one.

Q17. Stop-and-Wait Efficiency

A Stop-and-Wait ARQ efficiency question with approximately:

Tp = 10 ms
Tt = 1 ns

Q18. Subnetting

Given an IP address of the form:

192.x.x.x/26

Find the number of available hosts.

Q19. CFG Question

Consider the grammar:

S → TaT
T → aTb | bTa | a | ε

Select the string that belong to the language generated by this grammar.

Q20. DFA

Minimum number of states in a DFA accepting empty language over finite alphabets 

Q21. Regular Expressions

Let L(R) denote the language represented by regular expression R.

Statement P:

L(R) ∪ {ε} = L(R) always holds.

Statement Q:

L(R)·{ε} = L(R) always holds.

Determine which statements are true.

Q22. Pipeline Stall

Instruction I1 is:

LOAD R1

Instruction I2 is:

R3 ← R2 + R1

Assume:

  • Standard 5-stage pipeline
  • Operand forwarding allowed
  • For load instruction, forwarding happens from MEM stage and not EX stage

Find the number of stalls.

Q23. Seating Arrangement

Two distinct cars are available.

  • First car can seat 5 people
  • Second car can seat 4 people

There are 8 people total.

Find the number of ways to seat them such that:

  • every person sits
  • no car remains empty

Q24. Linked List

Time complexity of reversing a linked list iteratively.

Q25. Load Factor

Consider the load factor α\alphaα in hashing.

Which of the following statements are true?

  • α is always less than 1
  • As α increases, the number of collisions increases
  • As α increases, the number of collisions decreases
  • (One more option, not remembered exactly)
 

Q26. EMAT

Given:

  • TLB lookup time = 10 ns
  • Main memory access time = 100 ns
  • TLB hit ratio = 0.9

Find EMAT.

Q27. GCD Properties

Which of the following statements is false?

  • If gcd(a,b) ≠ 1 and gcd(b,c) ≠ 1, then gcd(a,c) ≠ 1
  • gcd(an,bn) = n × gcd(a,b)
  • gcd(ab,ac) = a × gcd(b,c)
  • gcd(a^n,b^n) = (gcd(a,b))^n
 

Q28. Asymptotic Analysis

Suppose a function f(n) is both:

  • little-o of g(n)
  • big-Omega of g(n)

Which of the following is true?

  • Always true
  • Sometimes true
  • Never true

Q29 : Simple Single level Paging question asking size of Page table

I think the values were 32 bit Virtual address, page size is 2^12 Bytes and Page table entry size is 4 Bytes

Q30. NP-Hard Problem

Given a graph G(V,E) which of the following problems is NP-hard?

  • Finding a cycle
  • Finding a cycle of length 6
  • Finding a path between s and r of at most length k.
  • Finding a path between s and r of at least length k.

 

 

 


335
335 views

I recently appeared for the MTech CSE (Research) interview at IIIT Delhi and wanted to share my experience.

There were two professors on the panel.

The interview started with a basic introduction. After that, the first professor asked me about the longest piece of code I had written during my BTech. I talked about my major project, and we had a discussion on it for around 4-5 minutes. The discussion went somewhat deep into the implementation and concepts. I was able to answer most of it, but I did stumble towards the end on a follow-up question.

Then the second professor took over and asked whether I had prepared specific subjects for the interview. I replied that I was comfortable with all core subjects. He then asked if he could ask a question from Discrete Mathematics, and I agreed.

Question asked:

Suppose you are given a set of size n. What is the number of subsets with even cardinality?

I answered: $$2^{n-1}$$

He then asked why.

I said that since the total number of subsets is $$2^n$$ , intuitively half of them should have even cardinality and half should have odd cardinality.

He then pushed further:
“Why exactly half?”

At that point, I paused. I hadn’t studied a formal proof for this before. He clarified that they didn’t expect a memorized proof but wanted to see my thought process and reasoning ability.

I tried to think along those lines but couldn’t arrive at a convincing explanation. He then asked if I was familiar with combinatorial bijection arguments. I said I wasn’t, and we moved on.

Overall Experience

The professors were very polite and created a comfortable environment throughout the interview. It felt more like a discussion than a grilling session.

However, I personally felt the interview was quite short and could have included more questions to better evaluate my knowledge.

Takeaways

  • Be prepared not just with answers, but also “why” and “how” behind them
  • Interviewers may test your thinking process, not just memorized concepts
  • Even if you don’t know a formal proof, try to build intuition on the spot

 


 

308
308 views

P.S This was an online interview

Panelists: P1: Prof Chandresh Maurya, P2: Prof Lawqueen Kanesh, P3: Didn’t speak

P1: Hello Introduce yourself.

Me: Good morning. My name is… I am from college… I have given GATE CSE this year and scored XXX. I have an interest in machine learning and have developed my final year project in that domain.

P1: Ok, preferred subjects?

Me: Probability and Linear Algebra

P1: What is Bayes theorem.

Me: We have an event E1 and A is the total probability of all the events. Then P(A/E1) ….

P1: Ok, what is expected value?

Me: Expected value is the statistical average value.

P1: If you toss a coin n times, what is the expected number of heads and tails?

Me: n/2 heads and n/2 tails.

P1: Prove it

Me: A coin has 2 sides, if we toss it then it either lands on heads or tails. When we toss it enough number of times the heads and tails converge to n/2.

P1: Tell me about linearly independent vectors.

Me: When we have a vector which can be formed by combining two other vectors then it is linearly dependent, if it cannot be formed by a linear combination of other vectors it is linearly independent vector.

P1: Do you know about spectral theory?

Me: No sir, I am not familiar.

P1: Ok then, what are eigenvalues?

Me: When we have a vector and upon applying a linearly transformation it stays in the span of the original matrix, then it is an eigenvector. And the value by which it scales, that scalar is called the eigenvalue for that matrix.

P1: Ok you said machine learning, if you have an inverted matrix … what is softmax?

Me: Sir I am still learning machine learning and am not very familiar with it.

P1: Ok then, let’s move to data structures. If you have 1 billion numbers and limited memory, what algorithm will you use to sort them?

Actually I encountered this question before and I remembered the answer being merge sort. But didn’t quite remember the logic.
Me: Sir merge sort.

P1: Why merge sort? Why not quick sort or heap sort?

Me: Sir quick sort has n² worst case time complexity and heap sort…

P1: Heap sort also has n² worst case time complexity?

Me: No sir, worst case is n log n.

P1: Ok, then why not randomized quick sort?

Me: Sir because we compute multiple iterations for finding the random pivot?

P1: Multiple iterations? Ok, anyone from the panel wants to ask questions?

P2: Tell me the definition of randomized quick sort.

Me: Quick sort, but we select the pivot element randomly.

P2: Ok, What is the expected value of the pivot element in randomized quick sort? Prove it

Me: Umm I am not very familiar. I have learnt from quick sort. //And I made up some things, mentioned about randomly selecting an element

P2: What is a random variable?

Me: A random variable x with respect to a probability distribution is, the element that when assign a value to the random value we can compute it’s probability for that value for that distribution.

P2: What is a binary search tree?

Me: It is a tree in which elements smaller than the root are in its left subtree and elements larger than the tree are in its right subtree.

P2: Ok, what is the time complexity of finding an element in a BST?

Me: Avg case or worst case?

P2: Avg case?

Me: log n

P2: Worst case?

Me: n²
I later on realised I made a mistake due to interview pressure/anxiety

P2: Ok we are done.

Me: Do you have any reviews on my performance?

P2: Revise the basics of algo, data structures and Discrete Math

Me: Thank you

Now look, the thing is that there were several panelists and people had different kinds of panelists. Some asked easy, some asked mid-tough questions like my panelists. But it all came down to luck.

Wishing you the best! I will also be sharing more of my experiences as well.


Follow me on my medium, I post regularly : https://medium.com/@thewittywizard

312
312 views

My GATE Score: 705
My rank: 1107

Honestly, the seniors of M Mgmt were very supportive. They made a whatsapp group before a week and gave us all instructions as well as advice on how to come and what documents to bring.

M Mgmt accepts CAT, GATE and GMAT scores. So around 300 people appeared for 20 seats.

My M Mgmt GD + PI was in the afternoon slot at 2:00 p.m. I had arrived early after my blunder at the Quantum Technology interview. Thankfully I talked to a few people and I felt light.

First there was a document verification queue and I was third in the queue so I was one of the earliest interviewees. There were two interview panels, one at the ground floor and one at the first floor. According to previous day’s data from a friend, the ground floor panel was math and statistics heavy while the first floor panel asked questions from your resume. I was on queue to the ground floor panel.

I was called in with the ring of a bell. My interviewers were Professor Chiranjit Mukhopadhyay[P1] and Professor Shashi Jain[P2]

I entered and greeted both P1 and P2.

P1: So, you are from CSE background? Introduce yourself?

Me: Sir I am a cse student pursing my btech from xyz college and I have an interest in management and related sciences. Sir should I also list my strengths and weaknesses?

P1: Smiled Oh we didn’t as that. But its ok. Tell me why you want to pursue a management degree?

Me: Sir actually I really admire the corporate structure and how it works. The intricacies behind it. How such a huge corporate works so swiftly despite having so many bumps along the way.

P1: So do you have corporate experience? or where is your view on corporate structure coming from?

Me: Sir I do not have any firsthand experience, but I hear a lot from my father who works in marketing in a growing company and from my uncle who works abroad in a huge corporate.

P1: So you do not have firsthand experience? I see.
Then he looks at the application form I had submitted
So you have done btech cse, you might have a probability course right?

Me: Yes sir.

P1: What did you study in that?

Me: Sir we learnt the basics of probability, random variables, distributions, total probability and in statistics a bit of hypothesis testing. I am not very familiar with hypothesis testing because it wasn’t in GATE.

P1: Ok, so you have two coins. One biased coin with heads on both sides and one unbiased coin with one head and one tail. Now you toss 5 times and you get 5 heads. What is the probability that its the biased coin in all 5 tosses.

Me: Sir we can solve this by Bayes’ theorem. I wrote down the probability if we tossed one coin.
But my nerves got the best of me and I couldn’t go further for 5 coins

But, professor Chiranjit guided me. He asked me what is P(A|B) I mentioned it is P(B|A) * P(A) / P(B) and also asked for total probability formula eventually writing it down and I arrived at the answer.
I even got so nervous I couldn’t calculate (1/2)/(33/64)

 

After this question got over, P1 resumed.

P1: So do you know hypothesis testing?

Me: Sir I am familiar with the central limit theorem.

P1: Ok, tell me about type 1 error.

Honestly, I didn’t remember much.
Me: Sir it is when the Null Hypothesis is wrong i.e reject and not accepted.

P1: So that is type 1 error.

Me: Sir I cannot really recall. But yes it is in the vicinity.

P1 & P2: Ok thank you. Any questions for us?

Me: Sir anything I could improve upon?

P1: No, no you were fine.

I walked out preparing for my GD.

I was called for my GD about an hour later. The GD constituted of 10 people. We were given a topic, on which we had to think for 5 minutes, write a paragraph on the topic for 5 minutes and have a group discussion for 15 minutes.

Seniors had advised us to be civil and if anybody had not gotten the opportunity to speak then let them speak on it.

My topic was “Emerging technologies are affecting mental health severely. Agree or disagree.” I disagreed and presented a few points.

The GD was being graded by Prof Anjula Gurtoo and Prof Sumirtha Gandhi. They were observing and writing down on a paper.

So that was all for my IISc experience. A tip from my side, if you come here then you will find cheap hotels near Majestic Bus stand. Its the center of the city and almost everything is nearby. IISc is about 6kms from here, and it took me 30 mins to get to IISc. Bangalore traffic is bad, so ALWAYS LEAVE EARLY.


Also, follow my medium because I post blogs there regularly : https://thewittywizard.medium.com/

364
364 views

My credentials:
GATE CSE Score: 705
AIR: 1107
Cut-off for Quantum Technology was somewhere around 600

So, I had my quantum technology interview yesterday. My biggest mistake was that I was hoping they will ask questions from subjects I am familiar with. Spoiler alert: They didn’t

https://www.youtube.com/playlist?list=PLNrdQx59gmEWV7eLbqDWeI_B72-0MXiyt

I suggest you go through this playlist. Btw the Prof Apoorva Patel sir, who is teaching in this playlist was one of my interviewers.

So, my interview was at 9:00 a.m. We had to reach the IQTI department. A few of us were sitting in the Instrumentation and Applied Physics department by mistake, but then we went to IQTI. P.S The location of IQTI is on google maps on their website so be sure to use that and not be lost like the eight of us. So, we arrived late at the IQTI department, at around 9:20 a.m. And there were about 10–12 people already sitting there. So, there were approximately 20 people for the morning slot. And I am guessing about the same number of people for the afternoon slot.

After sitting there, we eight discussed a lot of things and talked. Apparently they were all from different backgrounds. Some from electronics, cse, da and physics as well.

We were given a form to fill, it included questions like:
What other depts have you applied to?
What pref was quantum technologies?
If you had a chance to change the preference of quantum technolgies, what would you change it to?

My turn came at around 11:20. And I was called inside by prof. Akshay Naik. This was my first every offline interview so my heart was pounding so bad.

I entered, and there was a panel of 4 professors sitting from Left to Right : Prof Akshay Singh[P1], Prof Akshay Naik[P2], Prof Apoorva Patel[P3] and Prof 4(Sorry I cannot really recall properly who it was, and upon matching on the website I couldn’t find a familiar face). Prof Akshay Naik was smiling all the time and gave out a comfortable and positive vibe.

So I sat down, and we started.
P2: Hello Good morning. Why did you choose quantum technology?
Me: Sir I believe quantum computing is the future, etc, etc

P2: Ok, so what should we ask you. Since you are from a CSE background, what do you know about quantum computing?
This is where I made a blunder. I said:
Me: Sir, I have read on the basics of quantum computing for this interview.

P1: Ok, so what do you mean by read for this interview?

P4: I think he probably means he studied a bit for this interview, not out of interest.

This was a really big red flag
So I desperately tried to justify, a failed attempt honestly

Me: Sir, I have an interest in quantum computing, but I couldn’t find sufficient resources.

P2: What subjects should we ask you from?

Me: Sir I have studied linear algebra and probability and statistics.

P1: What subject related to quantum technology?

Me: Sir, matrices? linear algebra?

P2: Ok, go to the board and plot a graph for log x.

My anxious ass plotted the graph for e^x

P2: Ok, what is log (a*b)

 

Me: Sir it is log a + log b

P2: Ok, now what if a is 1 and b is e , What will log(a * b) be now

Me: Sir, log 1 will be 0 and log b will be 1.

P2: So what have you plotted there?
Ok, nevermind, now plot the graph for e^x

My anxious ass again plotted the graph for mod|x|

P2 & P1: Covid batch?

Me: I smiled a bit.

P2: What was the graph like during covid? How were the cases rising?

Me: Sir the patients were rising exponentially.

P1: So how was that?

Me: Sir exponential rate of growth means that the number of steps in every iteration are double the number of steps of the previous iteration.

P2: So, how was that graph?

I plotted it correctly now

P2: If e^x is like this, then how is log x related to e^x

Me: They are the inverse of each other.

P2: So you see your log x graph there? Would you like to correct it?

Me: Yes sir.

Professor Apoorva finally spoke
P3: If you are computer science student we atleast expect exponential and basics

I was shattered internally

P2: Ok thank you. You can leave now.

I blundered this interview so bad because of my nerves. Guys I really suggest you atleast practice some mock interviews before applying anywhere!
This interview(more like blunder) lasted around 12–15 minutes and then I left to get some food.

I also had a M Mgmt interview in the afternoon, and that went considerably better than this.

I will posting that soon.

 

Also, follow my medium because I post blogs there regularly : https://thewittywizard.medium.com/

948
948 views

IIT Kharagpur MS (Research) Admissions:

Most aspirants assume IIT Kharagpur follows the same direct application portal as IIT Madras or Bombay. However, KGP has a unique admission system where the journey often begins with a job rather than a student ID.

Step 1: Get Hired

To be eligible for an MS (by Research) at IIT KGP, you generally need to be working as a staff member on a sponsored research project within the institute.

  1. Find a Project: Monitor the IIT KGP SRIC Portal for "Temporary Jobs."

  2. Roles:

    • JRF/SRF: Requires a valid GATE score in a relevant branch.

    • Project Assistant/Associate: Often open to those without a GATE score (though GATE is always an advantage).

  3. Selection: You apply for a specific project. Selection is usually based on an interview conducted by the Principal Investigator (PI) and Co-PI(s) of that project.

Step 2: The Transition to MS

Once you are employed and working on the project, you become eligible for "MS Internal" admissions, which happen every six months (Autumn and Spring semesters). You will effectively be balancing project deliverables with your research and coursework. The CGPA and other criteria is subject to the academic and department.

Step 3: The MS Entrance Process

Even if you are already working at KGP, you must clear the formal admission cycle:

  • Written Test: Usually covers 3-4 core subjects (like Engg. Maths, Discrete Maths, Digital Logic, Programming and Data-Structures). You can find the specific syllabus for CSE on the Department Website under the MS (by Research) tab.

  • Interview: If you clear the written cutoff, you move to the technical interview. For MS candidates, this typically happens the day after the written test.

IIT Kharagpur PhD Admissions:
 
For PhD admissions at IIT Kharagpur, the process is more flexible but equally rigorous, allowing you to apply either as a regular Institute Scholar (based on a high GATE score) or as a Project Staff member (JRF/SRF). Unlike the MS route, which almost strictly requires being a project employee first, PhD aspirants can apply directly through the official admission portal when the call for applications opens each semester. The selection follows a tiered structure: you are first shortlisted based on your GATE score or academic credentials, followed by a departmental written test and a technical interview, which often on the same day. If you are already working on a project at KGP, you can apply to convert your position into a PhD, similar to the MS process, but the evaluation remains centered on your research potential and core subject depth.

Recently I have heard that the PhD Admissions will now not wait for 6 months cycle but was unable to find proper source for it, will update it here if I find so.