Vardaan Taneja, EE!
Interned at MIT Watson AI lab and SURA
Q: What did you work on in your SURA and Foreign Intern?
My SURA was under Professor Manan Suri. I had done a foreign internship with researchers from the MIT-IBM Watson AI Lab. My foreign internship was in the domain of robustness of AI methods, specifically in image reconstruction used in MRI. Through these methods, you could conduct an MRI, for instance, in less than 5 minutes. My SURA was related to neuromorphic computing, which models the way the brain works through spiking neural networks. I focused on these; they are dubbed as the next generation of AI. They could lead to significant reductions in modern-day AI’s energy and computational requirements. After all, you shouldn’t need 10 GPUs to do what your small brain does at 20W of power.
I had mailed a little in January, as I was unsure if I would be allotted a SURA project. I got around 12-13 replies, of which 5-6 were positive; some of these I received months later. I was already working with Prof Manan Suri during the semester on a project that was eventually converted into a SURA. I wasn’t particularly sure of research, so the broad goal was to go through the process of publishing a paper and explore what research is like. Once I was allotted my SURA, I mostly rejected all the offers from my foreign internship mails.
However, I eventually sat out of the OCS process on further contemplation. Moreover, I realised that it would take a long time for the SURA to result in a paper, which is why I mailed the MIT Watson AI lab professor if he still had a position available for me. I loved my time with them and witnessed how meaningful research can be done. I worked for 3 months beyond the internship period during the semester on the project and eventually submitted a paper as the first author at an IEEE conference.
I sort of got the flak for trying to do two internships simultaneously: peers and even parents told me I wouldn’t be able to do either of them properly. I actually had a close friend text me long paragraphs telling me not to do it. This is a common theme at IIT; people deem working more than necessary as redundant and ineffectual. I mean, I was doing two, not six or seven. Doing both was, in retrospect, a great idea. Not only did I get a paper out of my internship, but it also helped me build further perspective on the importance of ‘robustness’ and encouraged me to explore that idea in neuromorphic systems as well, which is now one of the central themes of my SURA work. I got to explore two entirely different approaches to AI in one summer. I don’t think I could have asked for more.
Q: What exactly inspired you to take up research and your opinions on research?
A: I have always wanted to start my own business, as I thought I could do more impactful work through that medium, and I probably still would try that at some stage. It’s hard to explore the path of running a company while you’re a student at IIT. Still, it’s much easier to explore research, so it seemed reasonable to spend some time doing that. The idea was simple: I wanted to work on real-world problems that I would be driven to explore and learn more about. Both research and entrepreneurship offer excellent means to do that.
I never really thought I was a ‘research’ person. I don’t know when it changed for me. I remember trying Leetcode and stuff for OCS, being bad at it and not liking it, which led me to think deeply about what I eventually wanted to do. This led me to read some excellent blog posts on 80000hours.org, which I would encourage everyone to go through. They document research on effective altruism, intractable or ignored problems, and try to derive conclusions about what sort of a meaningful career you can pursue in the future. I think that provided the perspective to explore research and let go of OCS for the time being.
I think people tend to be driven away from research due to a lack of interest in their major or the courses that they have done. However, research is vastly different. Almost all professors know that you would not have domain-specific knowledge and expect you to learn prerequisites as you go. I could try research in AI without having done a single ML course. This is also why some professors abroad would be willing to accept you in fields unrelated to your major.
Interestingly, our current SURA report has nothing to do with the initial proposal, which perfectly describes the essence of research: you don’t know what will or won’t work and at what stages you would make substantial progress. The learning was that you need to be persistent and original to do effective research instead of necessarily smart or hardworking.
Thoughts on foreign intern vs SURA and advice on the same:
I briefly struggled with that question, and I talked to some seniors. A foreign intern is not always the better option. On mailing, I realised that some internship opportunities abroad are about professors hiring you to do their labour for them instead of anything meaningful. I have heard about people flying to Japan to translate research papers. It’s not a bad opportunity, but it’s not research. In most cases, you’d be doing that for free if it’s remote. This does not mean there aren’t good opportunities, and I would encourage everyone to mail and see what is out there. However, in a SURA project, a professor at IIT has a vested interest in ensuring that you have a good research experience. That’s the whole point of the programme. Plus, you get paid a little. So I would encourage everyone to think deeply about what suits them better and not subscribe to existing notions.
As for the application process, I think we tend to obsess too much over our interests and the university we are applying to. I don’t think most of us know enough about any field to have well-defined interests yet. Further, our research experience is vastly more influenced by our advisor than the domain or school we are applying to, which we often don’t realise. So the focus should be on getting a good advisor, not a fancy school. I was fortunate enough to be advised by some excellent people, and I think that is primarily why I could be persistent and explore research properly.
Any other general advice?
I think the primary learning for me has been to never plan too much ahead into the future. The more I try to plan, the more my plans tend to fail. This keeps me open to new opportunities and what led me to try research. This is not the same as not thinking before I do things. If anything, I used to spend too much time doing and not enough time thinking about why I was doing it, which is counterproductive. I mailed a lot for internships in the first year because everyone else around me was doing the same, which led to a lot of stress and anxiety. That time could have been better spent doing something I was genuinely interested in and driven for instead of trying to fill up my CV. If you don’t like it, don’t do it. Vacations are relatively limited as it is, and you have all of the third and fourth-year to worry about the rest.
Interviewed by: Pratham Pahuja