Rice University – Tharun Medini – Univ Series

6 mins read
Start

Disclaimer: The content on this website is strictly the property of Insight, IIT Bombay. Content here cannot be reproduced, quoted or taken out of context without written permission from Insight. If you wish to reproduce any content herein, please contact us:
Chief Editors: Harsith Ravichandran (200260020@iitb.ac.in), Sanskriti Agrawal (20b030030@iitb.ac.in
)
Mail to: insight@iitb.ac.in

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

Q1. Introduction

I’m Tharun, 4th year Ph.D. student in ECE department at Rice University, Houston, TX. I work with Prof.Anshumali Shrivastava (https://www.cs.rice.edu/~as143/) in the RUSHLAB (http://rushlab.blogs.rice.edu/). My primary research area is to scale up Deep Learning algorithms to huge industry datasets using smart Hashing methods. 

As a part of my research, I interned as an Applied Scientist at Amazon Search in Palo Alto, CA for 14 months during 2018-2019.

Prior to joining Rice in Fall 2016, I worked as a Data Analyst at Target Corporation, Bengaluru for 1 year.

I did my B.Tech. in Electrical Engineering at IIT Bombay with a minor in Math.

Q2. What was your motivation for choosing a Ph.D. over a career in RnD

By the time I entered 4th year at IITB, I identified topics like Probability & Statistics, Image Processing and Computer Vision to be my interests. But I did not have an orientation for research yet. I worked on a couple of Computer Vision projects with Prof.Suyash Awate and Prof.Ajith Rajwade (who suggested that Rice has a great DSP group in ECE). I was a little indecisive about further studies and joined Target (a great place to work btw!). I got to read a lot of happening papers in Machine Learning and was amazed by the strides that the Vision field has taken thanks to Deep Learning. I wanted to be a part of this game-changer while it is still nascent. That’s how I applied for PhD at a few univs with strong CV/Deep Learning groups.

Q3. Did you have any specific motivation for choosing the US over some other places like in Asia, Australia, and Europe?

The abundance of good universities with my topics of interest, prospects of internships/jobs at the best companies and the huge Indian diaspora drove me to the US. My extended family has been in the US for a long time now which made my decision a no brainer.

Q4. Any exam tips, application tips, links to any personal blogs etc

In my opinion, a good SOP is the most important factor. Your SOP should reflect a motivating goal (can be a little ambitious) and what you did to get closer to that. Having at least one good Letter of Recommendation (LOR) is also important. It suggests how well you work with Professors (the essence of Ph.D.). Most students spend a significant time preparing for GRE/TOEFL while applying for Universities. But in my opinion, a GRE score might serve the purpose of filtering Ph.D. applications even before the committee of professors review them. I would say that a GRE score >315 should be good enough to get into most good universities. 

Q5. Any question that students should ask themselves before choosing a particular guide and institute for their PhD

Apart from the important questions like research topics, publication history of the group, citations, university etc., there are some other factors that you may have to explore. An advisor-student relationship is more than just academic. Your advisor bears some part of your aspirations and you’ll bear a part of your advisor’s aspirations. Hence, I would suggest applicants to email the students in a research group and ask about their research topics, how much facetime does their advisor give to the students, career prospects of students in terms of industry contacts and collaborations with other groups (for post doc opportunities/career in academia). Also, as students, it’s good to be flexible and adaptive to the vision of professors atleast for the first 3 years. 

Q6. What factors went into choosing your university, program, and guide?

I wanted to apply for universities with good vision groups. As mentioned earlier, IITB professors suggested about the great research in the DSP group at Rice. I applied with an aspiration to work with the computational photography group here. Rice has a unique process of admitting students. For the first 2 semesters, we focus largely on coursework required for Masters. In the 2nd semester, we are expected to work closely with a professor on a research project and then give a mini-defense presentation to a committee of 3 professors. This serves as a qualifier for further continuation of PhD. After this, we can continue working with the same professor or switch to another group based on mutual interest and available openings. I my case, I was exploring different groups in my first semester and multiple professors (including the stalwart Prof.Richard Baraniuk) suggested me to approach Prof.Anshumali Shrivastava (my current advisor) based on my interests. After having a couple of discussions with Prof.Anshu, I felt that his work is unique and super impactful for industry scale challenges. I worked with him for my qualifier and continued my association thereafter.

Q7. Differences between IITB and the current University in terms of Faculty, Facilities,  Research Opportunities

Rice is a relatively smaller university with a ~15 student-to-faculty ratio. Hence, undergrad students have more opportunities to work with grad students on good projects than any other public universities in the US (which are more like IITB with huge classes). That manifests in great research quotient for undergrads. Rice provides a huge shared computing cluster that most students find very beneficial to run demanding programs (IITB had such an option too but at a smaller scale). Many professors have their own GPU servers or Amazon Web Services credits which their students use. 

Rice has a great collaboration with Texas Medical Centre (right across the street) and hence many projects involve solving some crucial medical challenges. Rice and Baylor College of Medicine (pretty close to the university) have a great Neuroscience collaboration and usually, have many openings for integrating signal processing folks with traditional neuroscience folks.

Q8. How has your experience been, living in the US, and specifically, in your university town?

I’ve lived in Houston for 2 years and in the Bay Area for more than a year. My experience has gotten better with each passing year. I faced a severe mobility crunch in my first year because of not having a car. The public transport in the US is generally not comprehensive and may not cater to a lot of areas. Hence, going out for Indian stores/movies needed us to coordinate with friends who had cars. But that phase helped me get a bunch of great friends with whom I hang out very often. Now, I have a car and I do some payback for my juniors. 

People at the university at very friendly and make us feel belonged to, right from day 1. Houston is known for its diversity in food, wide roads (and we still have traffic jams) and hot and humid weather. Houston is a lot closer to India on several aspects except that rains suddenly gush out in all seasons. We have a vibrant Indian community (remember Howdy Modi!).

Q9. How is the social life in your university, and how is it different from IITB?

As a grad student, my social activity is far lesser than during my undergrad. To the extent that I see, Rice has a very good social life. We’ve several multi-national events happening round the clock in the center of the university which is open to all students. Unlike IITB, we don’t have any department outings to nearby hiking trails (Houston is a flat city). But we do have department Barbeques where people make burgers and hang out with other folks. We also have frequent sessions for job finding where we can meet influential people from the industry. There are seminars round the year with receptions where we can socialize with visiting professors and fellow students.

Q10. Some key takeaways from your Ph.D. experience.

Perseverance with an idea. Even the most principled idea doesn’t work right away. It needs careful assessment of steps to figure out how to get something working. Also, research is not about comparing with others and bothering about how many papers we publish. At the end of the day, I’ll be satisfied if my work has solved an important problem or has simplified an existing process thereby saving a lot of time/energy. Right now, I’m proud of the fact that we’ve excited a huge company like Amazon to use our algorithm which is simple and consumes much less time and computing resources for a better search experience.

Also, I’ve become much better and confident in social interactions. I’ve developed a good relationship with my advisor and fellow group members.

Q11. What are your plans for the future?

I’m fascinated by the thought of integrating large scale machine learning and bioinformatics and proposing efficient solutions to otherwise infeasible problems. I wish to apply for such research positions in places like Google AI/Google X and Microsoft Research. I also have some start-up aspirations with complimenting opinions from some lab-mates and my advisor. 

Q12. Any ideas for prospective Ph.D. students, and for fellow juniors pursuing their Ph.D. ( 1st, 2nd or 3rd year) if any?

Start early! I would advise undergrads to try co-authoring at least one paper with Masters/Ph.D. students. This gives a perspective of what it takes to conceive an idea and then present it in a comprehensive way with literature review, experiments, and baselines, plots, and tables, etc.

0

Don't Miss

New GRE – A sneak peek : Gaurav Lahoti

Disclaimer: The content on this website is strictly the property of Insight, IIT Bombay. Content here cannot be reproduced, quoted or taken out of context without

Student Gymkhana Awards 2015

Disclaimer: The content on this website is strictly the property of Insight, IIT Bombay. Content here cannot be reproduced, quoted or taken out of context without