Ritvik Gupta, MT1
Interned at: University of California, San Diego
I’m in the Mathematics and Computing department, so certainly, I had a leaning towards mathematics, per se. But I didn’t really want to pursue the abstract sides of mathematics any further, I specifically wanted to do something like applied mathematics combined with computer science. Then, as I was thinking over all this, the lockdown happened, and we had a break for about six months or so. I did a few independent projects and developed an understanding of machine learning through online sources like courses and research papers. By the end of that period, I was sure that I wanted to do an internship in Machine Learning.
Machine Learning draws a lot from probability and statistics, and I had a course on that in my third semester, which helped. Then, there’s the fact that Machine Learning courses are usually in the third year, so I had to learn a bit from online sources. I tried reading research papers too, but then some of them were hard to grasp. The main point is to just hang in there and try to get an idea of the stuff, which you will do eventually. Reading research papers also helped in the mailing process, as I mentioned them in the mails to the professors.
Overall, having a lot of prerequisite knowledge is not that important. What’s more important is a willingness to go beyond the needed stuff in your courses and venture online as and when the need arises.
I started mailing profs around the third week of November but stopped pretty quickly as I got a response on my fifth mail itself.
I think that the professors in Europe and the US, to whom you’ll be applying for a research internship, do not care that much about CGPA as much as we do here in IITs, so I don’t think that a so-called not-so-good CGPA will be a deterrent. Sure, a good CGPA is a plus point, but you can easily offset that by doing projects. In fact, I know several people who didn’t have a very great CGPA but still ended up with the most solid research internships.
So, I did my internship under a professor at the University of California, San Diego (UCSD) remotely due to the pandemic. Speaking technically, my internship entailed applying machine learning techniques inspired by human learning skills on differential neural architecture search methods. So, yeah, I specifically worked on the model of learning from mistakes, where we were trying to set up analogies between human learning and machine learning and to set a model for machines.
One thing I expected from a research intern was interaction with fellow research scholars and professors, but that was hampered due to the pandemic as I did my internship online. So that can be called a disappointment. But in terms of work done and knowledge gained, it was a very fulfilling and exciting journey.
Takeaways for juniors
Firstly, I think exaggerating is harmful, as the profs will know if you’re lying. They aren’t expecting a lot from you anyway, as research interns are for gaining knowledge and you are just in the second year, so lying can be easily caught and is unnecessary.
Secondly, I would say many people just do internships for the sake of it, as a part of a herd mentality. Meanwhile, others are very much interested in going deep into the research field. Like, there are so many things you can do, try for an industrial internship, work with a startup, and much more, so there are many things to explore. So, try to do something that interests you.
Interviewed by: Shrijit Shaswat and Tushar Srivastava