Reading Time: 5 Minutes

The content on this website is strictly the property of Insight and the Students’ Gymkhana IIT Bombay. If you wish to reproduce any content herein, please contact us:
Chief Editors: Saman Siddiqui and Varun Sule
Mail to: insight@iitb.ac.in

Introduction

Hi, my name is Kalpesh Krishna. I graduated from IIT Bombay in 2018, receiving a BTech in Electrical Engineering with a Minor in Computer Science.

I am currently a Ph.D. student at the Computer Science department at the University of Massachusetts Amherst, USA advised by Mohit Iyyer. I work primarily in NLP, or natural language processing (specifically, generative modelling) and dabble occasionally in deep learning security for NLP systems.

Motivation for going for an MS/Ph.D. over a job

During my undergraduate degree, I did two internships – a GSoC at Mozilla primarily on infrastructure and automation software engineering; a research internship at the TTI-Chicago / University of Chicago on speech recognition. I had also done some research in Computer Science with Prof. Preethi Jyothi as my bachelor’s thesis. 

Overall, I really enjoyed both internships and felt both career paths have their positives and negatives (I still believe so). I preferred a Ph.D. for the following reasons –

1) I really enjoyed reading new research; the flexibility and independence in the work, project direction, and daily agenda; the sense of ownership

2) I felt my field of study (NLP) was entering a golden age. There were lots of exciting, unsolved problems with solutions having a potentially high impact on research and industry.

3) My advisors (Prof. Preethi & mentors at TTIC) strongly encouraged me to pursue a career in research since they believed it was the more suitable career path for me.

4) Finally, I felt excited and ready to take on an interesting challenge straight from undergrad, with the hope that I will learn several technical and non-technical skills on the journey. I thought 21 is a fairly young age with minimum personal commitments, hence a good time to assess whether research is for me 🙂 In the very worst case if things don’t work out, I still have time to go and find an alternative career!

If the answer for the above question is interest in research, then why not jobs in RnD

Jobs in RnD vary significantly. Several jobs are heavily focused on engineering rather than research. Several companies have a strict criterion of using their RnD teams to perform research which improves their respective products. Only some RnD positions perform fundamental research at the frontiers in Computer Science, often having a Ph.D. as their eligibility criteria. A Ph.D. in computer science is a very good way to open up a spectrum of job opportunities, including industrial positions in fundamental research.

A question that the students should ask themselves before applying?

It’s good to see both sides of a coin via internships (academic research and jobs) before deciding what is more aligned with your life goals and interests. Read a lot of articles trying to understand what a Ph.D. degree is all about (for a list, refer to the blog below in 4). Life at IITB is quite stressful, a Ph.D. is going to be no different + for a longer time period. Make sure you understand this before signing up!

Why MS over Ph.D. or vice versa

Overall, an MS didn’t make much sense to me since I was primarily interested in doing full-time research. My advisors (Prof. Preethi and mentors at TTIC) had a very good suggestion regarding this – go for a Ph.D. if you get selected in a good group with a good advisor. I did apply to a few MS programs, but they were mostly to account for two scenarios – 1) I didn’t get a suitable Ph.D. advisor/offer 2) I start hating research after my bachelor’s thesis.

Luckily, my bachelor’s thesis went off well and I got offers from good researchers in my field. That explains my decision 🙂

Any exam tips, application tips, links to any personal blogs etc

I wrote a few blogs to document my experience,

  1. A comprehensive list of grad school application articles written by professors – http://martiansideofthemoon.github.io/2018/05/29/grad-resources.html
  2. My experience preparing for GRE and TOEFL – http://martiansideofthemoon.github.io/2017/12/07/gre-toefl-preparation.html

In addition, a few NLP Ph.D. students in the USA (including me) are working on an article and a survey about applying to Ph.D. programs in NLP. Stay tuned! 

Factors considered in choosing the university, program, and advisor 

This was a complicated decision-making process and I had a hard time comparing the five offers I was considering. I spoke to over 30 people (professors and Ph.D. students in NLP) over Skype trying to understand the pros and cons of each place/group/advisor. In the end, I felt UMass Amherst was the most aligned with my criteria. The following points helped me make my decision –

1) My prospective advisor was just starting as a professor and had done excellent research during his Ph.D. I believed he had a very interesting research agenda for the coming years. 

2) The NLP group at UMass is big, diverse and strong. I was quite impressed with the overall department strength in ML (it’s generally ranked in the top 10 schools for AI subfields) and their prior achievements in reinforcement learning, information retrieval, and information extraction. UMass also has excellent computational facilities with one of the largest GPU clusters in the USA.

3) I liked the idea of living in a pretty and remote college town, right in the heart of nature. I thought this would be a nice change from the city life I had grown up in and a good way to stay focused. Also, if I missed city life, Boston and New York City were only a few hours from Amherst!

Experience

Difference between IITB and the current univ 

  • Teaching

Teaching is quite similar, but probably IITB tends to be more theoretical in its courses. Also, the per-course content is generally higher in IITB (which can be both good and bad).

  • Facilities

A significant difference, especially in computational facilities. NLP is a resource-hungry field, often requiring a lot of hardware accelerators. I was restricted in the kind of projects I could do at IITB due to this reason.

  • Research

I would say faculty at both IITB and UMass are both quite talented research-wise. However, UMass tends to receive a lot more funding for research, facilitating ambitious research projects. A lot of computer science research is conducted in the USA, so research seminars and talks by famous researchers in the field are common. Interacting with the best researchers in the field is quite inspiring and leads to an excellent exchange of research ideas. Finally, since the USA has a strong software industry, there are a lot of exciting research internship opportunities to keep you busy in the summer.

  • Friends and Social Life

Nothing beats IITB hostel life. I miss Hostel 6 and my wingies!

Future plans

No idea! I do plan to eventually move back to India, but I don’t have anything specific in mind yet. This is quite dependent on how my Ph.D. goes as well. I am really excited about the growing research opportunities in India (Google Research, Wadhwani AI, improvements in IIT’s research programs).

Advice to students

Choosing what to do after graduating can be hard. Talk to a lot of people who are pursuing each career path you are considering. Make sure you critically weigh your options, analyzing the pros and cons of each decision. However, if you are unable to choose between two options, it’s likely that both are great options for you. Go with what your heart says and never be afraid to take a risk!