Adit Krishnan, an alumnus from the Computer Science Engineering department, is currently pursuing his Ph.D. at the University of Illinois, Urbana-Champaign. He shares with us his story of how he developed his interest in research and how he pursued it.
Q1. Can you tell us about your topic of research and its scope in the industry?
My primary areas of research include natural language processing and recommender systems. I mainly work in developing machine learning and deep learning approaches for problems that fall under the broad ambit of recommender systems and NLP. So if you are working in machine learning, irrespective of the research problem you are solving, your skill set is applicable in many industries, from industries like communication to finance and banking. Many industries will want to hire you because of this skill and solve the problems they face. My work mostly involves dealing with text. Let’s say users generate text on social media or consume content on social media or from other heterogeneous sources, even content like images and videos. So some of my work deals with profiling users based on their consumption patterns and then recommending content to them. I work in developing models to tackle this problem. Currently, I am interning in Electronic Arts(better known as EA among students), which is a major game developer and publisher, one of the biggest in fact. My work here is very different from what I do in my research. Here, I am developing artificial intelligence for games. More specifically, I am working on developing intelligent navigation systems where the learner essentially uses reinforcement learning to learn how to navigate complex terrains in different genres of a game including shooters and strategy games and many more. The idea is to use reinforcement learning to train to be able to navigate these different terrains, and it doesn’t always have to be foot navigation. For eg, in strategy games, it becomes more about what decisions you’ll take depending upon your enemies, allies etc. I develop artificial intelligence and agents that essentially learn to play the game by repeatedly playing among themselves or with other real players. I am quite familiar with database research as well. Here, topics like distributed systems and cloud computing are really hot right now as everyone wants to move their compute infrastructure on the cloud. For eg, EA now wants to have its game server on the cloud and have games that do most of the number crunching there and enable users to access them seamlessly on their mobile devices. The connection between my research on machine learning and this is that lot of the machine learning systems need to be deployed on the cloud so they are constantly in touch with the user data and they are able to push the content onto the user devices. So these are two broad areas where there is a lot of interest in the industry and I can speak for them because I am quite familiar about it.
( When I told Adit how games like FIFA were so popular among insti Junta, he pointed out an interesting fact that though FIFA is only one of the nearly 200 games that FIFA develops, it generates about 40 % of their revenue).
Q2. Can you please tell us something about your university and about how it is different from IIT Madras?
I’ll talk about the course structure and the research aspects. About courses, I’ll limit myself to the computer science department because I have not taken courses from the other departments here. One primary difference is that courses here(especially graduate level course) tend to be very heavy on the most recent state of the art research and that is something I found lacking in IIT Madras. What you’ll find here is that most of the professors are advertising their research through courses because it is a good platform to do so. The way students benefit is that they are exposed to the absolute state of the art, the cutting edge research that is going on, particularly on the database and data mining area. The professors spend a lot of time talking about the recent publications either directly or indirectly. This is in addition to the material covered in the course. This is something I found missing in IIT Madras. Here, the bar for research is set very high. People here are really aware of the recent work. I would like to point out that even at IIT Madras, the course structure is quite flexible when it comes to electives. In fact, there are some programs here where you are pretty constrained on the courses you can take. Applicants should read up about this before they choose a program.
Q3. Can you shed some light on your apping process? How did you decide you wanted to go into research and how did you put your app together?
The decision to go into research is not a spontaneous decision. It is something that you build up during your undergraduate course. Since I was in a dual degree program, I had to write my masters thesis anyway. So I worked with Prof. Sayan in the computer science department( who is currently at IIT-D). He was the one who really cultivated my interest in research because his course was heavy on research and we did a research project. I also did internships with IBM Research and Microsoft Research, which are a couple of good places in India for research. If you intern in such organizations and publish your work, they will strengthen your application as well as give skills to help you to succeed in research. If someone is interested in research, then in the third year it is advisable to apply to programs such as SN Bose, MITAACS etc. By contacting professors in such institutes, you not only land an internship but also potentially get a future advisor. For Ph.D. especially, it is the professor who chooses you, so if you know the professor, or worked with him or shown interest in his research it will go a long way in strengthening your application process. Other than this, focus on writing the statement of purpose(SOP). The standardized tests like SAT are not super important for the Ph.D. applications, but they do make a difference for the MS programs. Also, if you haven’t worked with the profs you want to apply for, you should make sure that their area of interest matches to your areas of expertise. You should mention these profs specifically in your application, that you want to work under them and tell why you will be a good bet.
You should get a Letter of Recommendations(LOR) from professors you have worked under, as they will be the people who can write about you in the letters. You can also take LOR’s from faculty members from whom you have taken multiple courses. If you have stood out in these courses/projects, then the professor can remember and write about you. You can take the third LOR from the industry person you have worked under. Typically there are three recommendations from profs that are required but if you can get more that is even better.
Q4. Can a person with a non-CS background interested in topics like ML end up apping into those programs?
Yes, of course. One important thing that applicants should keep in their mind is that even if you are not from a CS background, you will be considered in your area of interest, take ML for example. You need to focus on your interests in your statement of purpose and you need to focus on what you have done in ML. You need to talk about why you are well equipped to handle CS courses, for eg, about your programming background. You should show them tangible outcomes and the best outcomes you can point to are your publications in the area. Published research is much more valuable. It will also help immensely if you have interned somewhere or more if you have interned at any US university or other professors who are renowned in their field and this will be a strong indicator of your ability to succeed in the area.
Q5. How was your insti life like? Did you do any POR’s and how did these activities help you?
I did participate in Shaastra in my first two years but that didn’t have any significant impact on the research. POR’s help more from an industry angle as these help develop leadership skills. They do help on the whole too but they don’t have any immediate impact on your grad school app. I participated more in programming and other tech stuff. They too don’t have much impact on the app but if you have some nice things like progressing through some ACM ICPC rounds to mention then that is always good. Also, you should not just mention that you ‘did’ something, you should mention your achievements in that.
Q7. How is the life of a grad student in the US like? What do you generally do apart from academics?
There is nothing concrete as such. It depends on where you are studying. You should approach research just like a job. You should have clearly demarcated work hours as it is very easy to lose track. In research, there is nobody reminding of what has to be done. This is one of the differences between IIT and here. A lot of the predeveloped structure is removed. You do have periodic meetings with your advisor but you are free to decide on your activities. So you have to manage your time.
Q8. Last but the most important, what advice would you like to give to your juniors. Not only those who are apping but insti junta in general?
There are two main things that differentiate IIT from other colleges. The first thing is the fantastic faculty. You should exploit this to the fullest. You should interact with the faculty more often and this will good for both students and the faculty as they will gain students who can conduct research along with them. Second, choose courses that really add value. While grades are really important, you should explore some tough courses and not end up taking all easy electives.
Adit also added that any student who needs help regarding the apping process can directly contact him at firstname.lastname@example.org
Author: Sriram(BT-CH ’21)