Branch & Batch: 2018 / Electrical & Electronics
Masters: MS in Robotics at CMU
Thesis: Advanced Robotics Centre, NUS Singapore
CGPA: 8.88
GRE: 330
IELTS: 8.5
Q. Can you briefly introduce yourself and talk about what you are doing currently?
I'm from the 2018 batch and majored in Electrical and Electronics. Somewhere down the line, in my third year, I realized that I enjoyed the fields of Machine learning & Computer vision and decided to pursue them. I spent most of my third and fourth years working on them. After graduating, I am now doing my Masters in Robotics at Carnegie Mellon. For my masters, I am working at the AIR-Lab in the intersection of Robotics, 3D vision, and Rendering.
Q. They say getting your first few projects is relatively harder, how did you go about getting yours?
When I started with Phoenix in my 1-1 and joined the tech team, we planned to build a line follower robot for which we had to make a 'smart glass.' I just knocked on a Professor’s door whom I liked and asked her if she could mentor me in this project. While I was ready to develop the scheme and work plan by myself, she thankfully agreed, and I managed to secure funding for this project. And that is where my official journey in Robotics and mapping started.
After that, I kept reaching out to other professors in my 2-1 and 2-2, After which I started working with Prof. Parikshit. Under him, I was working in the field of Semiconductors and devices. As it was new for me, I asked him if I could work informally initially, and thankfully, he was open to the idea. Eventually, I started working for him formally, from my 2-2 to my 3-1.
About getting your first project, I'd suggest just going to professors and asking them in person. They are much more likely to give you a project this way than by cold mailing them. Reaching out to Professors through clubs is also a good idea. You can alternatively also reach out to seniors who would be able to mentor you.
Q. Can you tell us about the projects and internships you have taken in BITS?
My first project was the line follower robot, which I did in my first year. After that, using the 'Student innovation fund,' we tried to create smart glasses for disabled people. Although this project did not quite take off, I learned many things from this experience, like ‘Robot localization and mapping' and ray-tracing concepts.
During my time at IEEE, Aero club and Hyperloop India team, I worked on quite a few projects that involved mobile robots, and Vertical takeoff-landing aircraft. We managed to publish a paper on the aircraft project in my second year.
I also had an opportunity to intern in my second year at a startup in London which was developing autonomous robots for indoor environments.
In my 3-2, I had my formal projects in healthcare and AI, after which I worked on my thesis in my fourth year, which was on neural rendering.
Q. You said you worked on many self-projects. How hard is it to come up with ideas for them?
Initially, it was pretty tough. I just brainstormed concrete ideas which I thought I could implement. After that, I just read many papers to familiarize myself with current research and essential topics in computer vision and robotics.
You just have to be patient. It took me almost 6-7 months to start coming up with good ideas. And even when I did get an idea, I used to find that it was already published and felt frustrating . Eventually, because you keep reading more papers, your ideas keep improving and get more refined.
When I had a base idea, I just googled it to see if it was even possible. I just kept going through different papers on that idea. I started thinking if a particular idea on one paper could be adopted in a different way somewhere else. As I got more experience, I could start combining various fields on a single project.
Q. What resources at BITS or outside helped you pursue the field of Robotics?
I would say that my first stepping stone was the Deep learning specialization course at Coursera. From the beginning, I told my professors that I was interested in robotics and computer vision, and I asked them to give me projects aligned with this field. Fortunately, they were open to considering my request which let me explore it further.
Apart from that, I had also taken the Robotics and Automation minor, which did help me learn a few things, but it was not that helpful as I already had a basic understanding of the field by my 3rd year. Nevertheless, the minor is a good start for a beginner.
My thesis and research internship contributed to most of my learning as you build skills much faster while working on it.
Q. Did you consider sitting for campus placements and summer internships?
I did not actively sit for them, though I did register. In my 3-1, I tried a few companies for summer internships, but it did not work out for some reason or another.
I got placed in a robotics company that came for on-campus placements . But I didn't go down that path. I also applied to companies off campus to ensure I had backups and interned with a startup in my 4-2. They extended me an offer for a machine learning and computer vision engineer. Ultimately, I did not go through with that once my admission came through. To whoever is applying for grad-school, I would say it is not prudent to carry on without backups.
Q. How was your experience converting from PS to a thesis?
I started cold-emailing professors in my 3-1 around September and October. I believe it's always better to go for quality over quantity.
Rather than emailing a hundred different professors with the same email, it is better to find a couple of professors that fit your interest.
Don't keep the email too long. Instead of listing your entire CV, discuss how you and the Professor's lab would be a good fit. Also it's good to reach out to contacts and PhD students of that lab.
I got my first acceptance by the 15th -16th mail. The acceptance email for my NUS thesis came towards the end of October. Then I finally got my formal letter in December. After that, the paperwork of converting from PS to thesis was not that hard and was relatively straightforward.
Q. You were supposed to do your thesis for one semester. How did you extend it for one whole year?
Officially on paper, my thesis was only for a semester that was from August to December, but then again, you can start it early if you want, nobody is going to stop you. I reached out to my professor in February and said that I'm ready to work right from then so as to get a better idea of what I'll be working on by August. As an added positive, he was quite happy with this. I soon got assigned to a PhD student. The formal dates of the thesis will definitely be there, but you can always start early, that's usually not an issue.
Q. Coming to the masters, could you explain the process of the exams you had to write and the universities you applied to?
I gave the GRE in August and IELTS in September.The universities that I applied for my MS were UCSD, University of Michigan Ann Arbor, Arizona state university, Georgia tech, and Carnegie Mellon, most of them in ECE and CS departments. In Canada I applied to the University of Waterloo, I applied to two departments there, systems design and electrical. I also applied to KTH Sweden and the Australian National university.
Q. Did you apply to any German universities?
Yeah, Germany has free education, and I was actually quite keen to apply to the EU in general. Unfortunately, most of my work is in computer vision and machine learning and except for one odd university with an electrical department that works in machine learning, the rest of them are just CS schools that I was applying to. The EU is very strict in this matter and they want you to have a set of specific courses done, they officially ask for your handout to check if your courses are on par with their standards. And since I had not done quite a few core CS courses, there was no point in applying, and just to give you some perspective, a few CS people also get rejected in Germany because our courses are at times considered to be ‘inferior’ to theirs.
Q. Did you consider doing masters directly from India? What are some of the things to keep in mind if pursuing masters from here?
I didn't consider applying for universities in India. You have some good universities here and should consider them if you're willing to give GATE . Of course, the funding may not be as great and research facilities may not be as good as big labs outside. Outside, you usually don't have resource constraints that you face in India.
Q. What will you advise your juniors while preparing their SOP?
The first thing is not to leave it for the last minute. It takes a lot of time to craft a good SOP and definitely don’t blindly follow a googled template. The admission committee and professors receive these applications every year and so are extremely well experienced to filter them out. So you have to first analyze why exactly you want to do the course. You have to answer a couple of important questions :
Why exactly you want to do this course specifically?
What experience have you had before this?
Once you do enter the course, what are you planning to work on?
How will this help you in your career goals?
You don't have to butter up the pros or anything, you have to give them valid points. Once you have your first draft ready, you can either approach your seniors or your peers who are applying that year for them to proof-read it and give you suggestions. You can also approach your parents and friends to get their opinions. It took me almost a month to have my SOP ready.
You should also change your SOP for each university because for example the research being done at Stanford is not necessarily what is being done at Carnegie Mellon. Some of the other things you have to keep in mind are to mention professors' names and the research that interests you. Answering the aforementioned questions should leave you with a good foundation and you essentially have to tweak a paragraph or two every time you change a university.
Q. CGPA, Research projects, LORs, SOP : How would you rank the importance of these in building one’s profile?
Research is the most important factor. If you've done good research, you'll be able to talk about it in your SOP and it improves your resume. Your LORs will automatically be really good as well. I’d say first is research, second LORs, and third your SOP.
Try maintaining a decent score in IELTS and GRE. It doesn't really matter if you have a 330 or 340, but I would advise not to have anything less than 320 if you're applying to the top schools.
These are the four main priorities. With respect to CG, the more it is the better. If you have a 9+ CG, it's just going to make your life easier. At the time of applying my CG was 8.6, but I've seen people getting in with an 8 (and lesser) too. So CG is something you should try to maintain throughout your life at BITS because it'll help you later on, but your CG should be maintained parallelly with good research experience, LORs ,SOP, and test scores.
Q. What publications do you have and are the publications necessary for getting into the top tier Universities?
I have in total 6 publications as of now. These were definitely helpful while I was applying for my masters, but then again, just publishing should not be your main aim because if the quality of a publication is not that great, or if it's not really relevant to what you're applying for, then there is no use.
Most of my publications were in AI for healthcare, and I got a Robotics and Computer Vision admit. I talked about the research I did for my publications in my applications to highlight the fact that I was committed towards research. But what really got me admitted was probably my thesis and that is not yet published, but I told them what I'm doing and I submitted my thesis document during the application. So it's perfectly fine if you don't have publications as long as you can talk about your work and your professor can vouch that you've done proper research and are planning to submit to some good journals. Having research experience is necessary for getting a good top school admit, and while it’s perfectly fine if you don't have any publications, having relevant ones in good journals would definitely help.
Q. What do you think is more useful - the number of publications you have or having publications that have been cited more?
I would go with the latter. At least in computer vision, what usually happens is the field tends to move very fast and if your work is getting cited, it means that people consider your work is still relevant even with the changing times. But then again if you are a fourth or fifth author in that paper, then it’s not very useful to you, because they consider your position in the list of names.
I would prefer having a lesser cited paper as long as you are the first or second author. Being the first or second author shows that you are capable of doing independent research because these positions are generally given when you have spearheaded your entire research project. Of-course there may be some cases like dataset papers where there is a large list of collaborators and its fine to not be a leading author there. For projects that you work on under professors, they usually take the position of last authors. If you feel that you and your collaborator have done equal work, you might just be named second in the paper, but in the footnotes you are told that the first and second author are equal authors. So that makes you a co-author of the paper, which is completely fine as well.
Q. How does the tier of the journal impact the importance of your publication?
If you're publishing, try to publish in the topmost tier. For computer vision, you have CVPR, ECCV, ICCV, IJCAI, NeurIPS, these are some of the top places, the leaders in that field publish their papers there and whatever is published there kind of trickles down at the bottom. So if you have a top tier publication, that'll almost guarantee you an admit in a university, provided everything else is good enough.
In journals, you should look at the impact factor and this varies across the field. For machine learning and computer vision, IEEE PAMI is the top tier. It has a very high impact factor, and it's very selective. But if your paper does not get accepted, then you have good mid-tier journals like Neurocomputing, Elsevier. That is quite good, and it's quite reputed. I would say don't try to go below that, that won't really add to your application.
Q. When did you start working on your first paper and how was your experience doing so?
While the idea kind of happened all the way back in 1-2, the work properly started a little bit later. I definitely had to pull some all-nighters while working on it. Writing a paper is painful in its own way because the fun part, the technical work itself, is actually the easier half of the equation. The procedure that follows afterwards is really dry, not fun and you can’t wait for it to get over.
Q. What is the difference between getting your paper published in a journal and presenting it at a conference?
Definitely procedures. The main advantage of a conference is that it does not take as much time as a journal’s revision. You have tons of revisions in a journal submission which will go on for anywhere between 6 to 8 months. Whereas at a conference, review time will be very short, about two to three months and you'll get a decision in one go.
This could also be seen as an advantage in favor of journal submissions though, since the multiple rounds of revisions mean that you are able to rebut multiple times. If a reviewer at the conference rejects your paper, your paper is rejected for that conference, and you either apply to another conference or a journal. And generally publishing a journal paper is more prestigious than a conference paper because it takes more time and the vetting process is much more rigorous. But if you have an A+ conference under your belt, that's almost as good as publishing in a journal.
Q. Which technical clubs were you a part of in BITS and how did they help you?
I was in almost all the technical clubs related to electronics, my first club was Phoenix. I was in ARC till 2-1. I was also a part of Aero club and IEEE. I was the Treasurer of I-Cell, but then COVID happened. I was a part of SEDS for a while. These are the primary clubs that work in the field of robotics and ML, each of them had some importance in improving my profile. During my time at Phoenix, I learned to think independently, read papers, and research independently. At ARC I learned more about swarm robotics and my peers were good as well. So the learning process was always good. Unfortunately I had to leave in 2-1 because my responsibilities in other clubs were quite demanding. I was the IEEE Joint-Sec and some other things. Aero club was probably a big factor for my applications because that was the place I got my first paper from. I think IEEE was a good learning experience because the club was not exactly active in my first year and we had to work quite a lot to make it active again.
Q. How did being IEEE chair help you, apart from the technical aspect, how did working with the members benefit you?
Being the IEEE chair was purely a managerial thing. I had quite a few learning lessons there, including planning out what the club was going to do for the year. It might seem like a simple task, but you have a lot of people with you in the club and not everyone's time is going to align with your initial requirements, so planning around that was quite the challenge.
And COVID made the year more difficult than it had to be, but I learned how to start a club from scratch, take new people, make sure that there are projects that keep on going and create a structure that is self-sustaining.
Q. You were a part of the tennis team and the athletics team as well. How did you manage that alongside academics and technical clubs?
I don’t think I managed academics. My CG was not that good until my third year, during my second year I wasn't doing that great in my CDCs. I used to always manage an 8, but not many 9s or 10s, my CG was always hovering about 8.2. That was just a day or a week of studying. I never really attended classes except for some that I really liked, pattern recognition was a course that I had done in 2-2 and I attended every single class for that. In your third year you have less CDCs, you have DELs. So when I took the DELs of my interest, it kind of just started coming naturally. So that wasn't a big effort.
Q. Do you have any advice for your juniors? And do you have any regrets?
CG is something that should be constantly maintained. You can't afford to let it go below a certain extent, try to maximize your CG as much as possible. Coming to the advice part of it, in your first year, try to explore all things. I have some regrets that I didn't explore things like finance or DSA or even be a part of cultural clubs. I think in your first year you're free, so explore everything, see what you like.
In your second year you should have at least a broad idea. Say there are two fields that you are interested in, start diving more into it, ask your professors for projects. But remember to maintain your CG at the same time.
Third year is generally the most hectic. You have to have a good CG, because those will be your final semesters when you submit your application. So if it goes downhill there, then your admissions committee might ask you about it. And you also have to be serious about one area. For me, I realized that that area was machine learning in my third year. So I had to start off from scratch and do a lot of work from the ground up. I only had six to eight months to build my profile. So to kind of ease it up, just have a better idea already by the end of your second year.
The fourth year you just continue to work on your thesis, you can reach out for more collaborations now. You won't have much academic load.
I think that is the yearwise breakdown.
As far as regrets are concerned, I think that I should have maintained my CG more during my 1-2 and second year, that would've obviously helped a lot. And I did not really enjoy college that much until my fourth year. So I definitely do regret that, but again, sometimes there's a trade off. You have to see what you want more.
Q. Is there anything more that you would like to add?
People often get bogged down by the administrative formalities, they think it's too late to do something and regret not starting things earlier. But if you really want to do something, then there are a hundred different ways of achieving that, you just have to be willing to be persistent. Also, always have backups to all of your plans because you're never too certain of what's going to happen until you have the end results in your hand.
Disclaimer: The points given above are the views and steps taken by the individual. They are not fixed steps and guidelines to base your college upon. Our hope is to inspire students so they can take the necessary steps hereafter. We hope you like it!
Comments