Univ Series: Carnegie Mellon University

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: Shreerang Javadekar, Shreeyesh Menon
Mail to: insight@iitb.ac.in

Deepak Dilipkumar (DDK) is currently enrolled in the MS programme of the Machine Learning Department of Carnegie Mellon University.
Vipul Singh (VS) completed his MS in computer Science at CMU, and is currently working at Bloomberg in New York City.

What made you choose academia over taking a job? Or any other career choice?

DDK: I haven’t really chosen academia as a career path yet (I’m not currently planning to do a PhD or become a professor), so I’ll talk about why grad school over other things.
It was a last-minute decision on my part to apply to grad school in Machine Learning. I was never really planning to go into core Mech, and I spent most of my second year and third year believing that I’d be going for a non-core job, or an MBA. Just before I went for my third year intern, I was pretty much fixated on giving GMAT and applying for an MBA abroad the following semester.
But the intern (which was at a certain finance firm that I shall not name), was a bit disillusioning. I realized that the kind of corporate atmosphere I’d be exposed to in a traditional Finance or Consulting role was not one that I enjoyed being in. Coincidentally, it was around this time that I started thinking about ML as a career path, mainly due to a blog post I’d read about AI (a fact that I mentioned on my SoP and got plenty of criticism for from reviewers. ¯\_(ツ)_/¯ ).
I spent a lot of my 7th semester reading about and working on ML, and realized that this was definitely the sort of thing I wanted to be doing. I didn’t have the necessary skills to get into a core ML role however – partly because I didn’t even have a formal CS or Stats background. So going to grad school to pick up those skills was a pretty easy choice after I had decided that ML was what I wanted to be working on.

VS: I knew that I wanted to specialize in some field, and spend some more time in academia. I had interned at Facebook HQ during the summer before placement season and realized that I wouldn’t get to do work that I wanted, with just an undergraduate degree.

How are you handling the finances? What are the average monthly expenses and inflows (if any, from stipends, TAship, etc)? Is the financial situation as bad as portrayed in the general consciousness? Some hard numbers (if you are willing) would be helpful here.

DDK: CMU is expensive. I mean, really expensive. The official numbers on the website suggest around $60,000 dollars per year. The monthly expenses come out to around $1000 including rent and food and miscellaneous stuff to buy (like the odd PS3 once in a while), but it’s the tuition that’s the real problem. $21,000 per semester to be exact. I have an RAship that covers my monthly living expenses, but doesn’t put a dent on tuition, so I’ll end up spending around $60k out of pocket across the 3 semesters that I’ll be here. A bit of a step-up from IIT. A well-paying internship could bring that down by another 15-20k, but it’s still a lot.

So tl;dr – yes, a Master’s programme in most American universities (and specifically in CMU) is expensive.

VS: During the three semesters I spent at CMU, my overall expenses (tuition, rent, food, etc.) came to around 80k$. I had saved around 15k$ on tuition fees by doing extra courses in my first two semesters. You only pay for 36 credits a semester, so if you can handle a higher course-load, you can get away with lesser requirements in the last semester, and hence save a lot. I was lucky to be financially strong enough, thanks to the internships and all the post-JEE scholarships/rewards. I was a grader for an undergraduate course during my first two semesters, but that hardly paid enough to cover the month’s rent, which would be in the range of 400-600$ for Pittsburgh. A research-assistantship in the final semester paid me much better though, with around 1000$/month.

What are some key tips for apping that you learnt about during the apping process that you wish someone had told you earlier?

DDK: Start on your SoP early. From my experience, GRE, TOEFL, recos and resumes get finished on time one way or another. The SoP requires a lot more introspection than most people realize. Putting all of your thoughts and goals and beliefs and motivations into a 1000 word summary is not easy by any means.

VS: I had applied to the PhD programs at UCB and MIT, and to the MS programs at Stanford and CMU. I guess I could have taken the risk of applying to a PhD program at CMU too. I think I associated a lot more stigma than there really is, to dropping out with an MS after joining for a PhD. At the time of applying, I was more interested in a different field, and didn’t find prospective advisors for a PhD at CMU. During my MS, I ended up specializing in a different field though.

Can you tell us about the job/hiring market currently (or projected for the next three-four years) in Industry or Academia where you are currently?

DDK: Machine Learning is growing rapidly. It’s a young field by most standards, but it’s really taken off in the last few years with the availability of huge amounts of data and the ability to perform heavy computations using GPU’s. Most companies that produce data of any kind (which nowadays is basically every company) have dedicated Data Science/ Analytics roles. Many of the companies that affect our lives on a daily basis, like Google, Facebook, Uber, Amazon, Microsoft and Twitter have dedicated ML roles, since for many of them, Machine Learning isn’t just a way to harness data – it’s their core product. So there are plenty of opportunities in both startups and established companies for people with strong backgrounds in the field. Joining research labs in these companies often requires a PhD in ML however.

VS: I currently work with the Machine Learning R&D Team at Bloomberg in New York City. I interned with them during the MS, and loved it a lot. The job market these days is ripe with opportunities for people in this field.

How did you choose (or how are you going to go about choosing) your advisors? What are the most important things to keep in mind while doing this? Are there any quick and dirty tips you could share with us?

DDK: I’d say that at the end of the day, it comes down to one thing – matching research interests. The academic atmosphere in the US is pretty different from what we’re used to in India. In India, there tends to be an invisible barrier between professors and students, even if you’re working under them. The relation here is a lot more informal and relaxed, which I think can only be a good thing. Considering that, you don’t need to worry too much about whether you’ll be able to get along with your professor or things of that sort – you probably will, irrespective of professor. So then the only factor when choosing your advisor is whether or not your research interests match well with the projects your professor is currently working on.

VS: After spending a semester at CMU, I thought of venturing into research. So, I went over the profiles of professors, and chatted with some graduate students I had made friends with. Another e-mail to a professor whose work looked interesting to me, and I was off to working with an advisor. Things to keep in mind – talk to people, make friends, don’t pester the profs but it’s ok to shoot them an e-mail and ask if you could meet.

What made you choose US over Europe when it came to deciding where to pursue higher studies? What are similarities/differences and pros/cons between the two?

DDK: The choice for me was primarily based on the state of ML research in different universities, and the presence of a terminal master’s degree with an emphasis on machine learning. I found universities satisfying both criteria mostly in the US. And at the end of the day, CMU was an easy choice considering that it has it’s own ML department. However, I think that what I mentioned earlier about informal relations with professors holds for Europe as well as the US.

VS: Having interned at IST Austria in summer 2012, I had seen how good they were at certain fields in CS, but lacked the diversity of research areas that is available at the top American universities. Combine that with the dominance of US universities in the rankings and the abundance of jobs here, and it was an easy decision to pick US over Europe. There are lots of great academic institutions in Europe too, such as EPFL, ETH, Max Planck, etc. and one should definitely consider them if one finds people or groups working in their areas of interest.

What made you choose to opt for an MS rather than to go directly for a PhD?

DDK: A PhD is a big commitment – both in terms of time and effort. Opting for a PhD is not a decision to be taken lightly. I was exposed to very little research at IITB (one of my main regrets from my stay) and I had been involved in Machine Learning for about one semester when I applied for grad school. I didn’t have nearly enough information about either the field or my capacity for research to make a 5-year commitment. And even if I was willing to do that, I didn’t have any research publications or even a CS background. Getting into a good PhD programme would’ve been a long shot to say the least.
Like I mentioned earlier, my motivation to go to grad school was to pick up the skills needed to work on interesting ML problems in the industry. I believe that a Master’s degree in ML is enough to open the door for these opportunities, and my current inclination is to go into industry like I had wanted earlier as opposed to continuing on to a PhD. This is true for a lot of people- An MS is a couple of years to pick up additional skills and give you time to think about what you want to do in the field.

VS: I had applied to just the top 4 universities, for two reasons – (a) I was pretty sure I would get into at least one of them, and (b) I wasn’t keen on going anywhere else. Once I was done with the application process, the choice was very simple. I got accepted only at CMU, so didn’t have anything to choose from.

What would you say has been the toughest lifestyle changes you have had to face? And how have you overcome them?

DDK: A room twice as big as the ones in H2, the freedom to choose the food I’m eating everyday, and a TV that doesn’t need to be shared by 400 people. I’m not complaining.
But as far as daily activities are concerned, acads do tend to take up a significantly larger percentage of my time that in IIT. The emphasis here is on assignments instead of tests, which gives you more opportunities to learn but has the disadvantage of not being doable with an all-nighter. Again, this is not really a tough change so to speak. If you’ve chosen to come to grad school, odds are that you’re genuinely interested in your field. So the assignments don’t feel like a burden; you just have less leisure time (or PoR time) than you’re used to.

VS: Most changes I had to face were welcome changes, I’d say. One thing that most international students find tough in the initial few months is that things are much more expensive if you start converting dollar prices to native currency. Not much you can do about it though, gotta live with it, and get used to it.

Is forming a new social circle in an alien country difficult? How lonely or not is life? A brief overview of life as a student would help.

DDK: Not much harder than forming a new social group at home. Sure, everyone comes from different backgrounds, but most of the people that you meet on a day to day basis tend to have a lot in common with you. You still live a student life essentially, and that comes with the activity and bustle that we’ve grown used to in insti. So no, life is not lonely (quite the opposite).
That being said, the major difference is the lack of wing/hostel culture. Whether or not hostel culture is dying out, being in contact with your batchmates 24/7 for 4 years is a very different experience from meeting them mainly in classes. You end up making some great friends in grad school, no doubt, but it’s not the same type of bond you tend to have with wingies and batchmates in IIT.

VS: This part turned out to be quite easy for me. Soon after landing in Pittsburgh, I started going out to welcome events hosted by local families, and made lots of friends, many of whom I am still in touch with. I would say that this varies largely from place to place, and from experience, is something that won’t happen in the Bay Area. Pittsburgh on the other hand is a very warm-hearted, friendly and welcoming city. Life as a student did get very intense and tough during the second semester, mainly because I went for 54 credits, including a research project, when the recommended load was 36. I used to spend 12-14 hours a day at my lab-desk, and soon had to make a decision to take Saturday off every week, something I still stick to. I would say that students should find some similar ways to relive themselves of the inevitable stress.

What are the career options someone in your place has right now, and why would you choose what you choose?

DDK: It basically boils down to: continue on to a PhD, or move into industry. From a PhD, you can stay in academia or join a research lab in industry. I believe that ML is niche enough that a specialized Master’s is enough to get me into a good industry role, which is why I’m likely to go into industry directly without a PhD.

VS: The American market is a great place. And as Mr President-elect says, it’s going to become great again. It’s an open market, unlike the closed placement-cell at IITs. This place really really rewards talent. By the time I had finished my stint at CMU, I had held internship offers from Microsoft, IBM and Bloomberg, and job offers from Google, Amazon, Oracle, and Bloomberg. I chose Bloomberg for the internship because they had promised me a research project in the field of Machine Learning, and I wanted to be in New York City when Roger Federer came to play the US Open 2015. I chose them for the job because once all the major players had leveled the field in terms of compensation and perks, I knew exactly what I was getting into, and I had loved the internship.