As a Ph.D. student, your job is multi-faceted. At any given time, your time is divided between: classroom study, Research , writing, “networking”, teaching, applying for fellowships, curating your online identity / promoting your online presence, reviewing papers, serving on committees, and wondering why you can’t do any Research. There are probably a few other things that I forgot, even.
It’s easy to forget about any one of these tasks . However, as a computer science graduate student, I think it is especially easy to forget about “applying for fellowships”. Why? Because we are living in an incredibly fortunate time. There is a lot of money in computer science research that comes from both government and industry , so your adviser can probably afford to pay for you even without any external support.
Still, I’ve applied for and received three fellowships throughout graduate school and have come close to winning two others . I may receive another before I graduate. These fellowships have covered almost all of my expenses as a Ph.D. student: tuition, stipend, student fees, health insurance. Cumulatively, I estimate their value, thus far, to be around $350,000. But, as I said before, none of these fellowships were strictly “necessary”. If I hadn’t had any of these fellowships, I am fortunate to be enrolled in a department that would have taken on all of these expenses without hesitation .
So, why apply for fellowships at all? In this and the next post, I’m going to try and break it down. But the basic point is simple: if you are currently a Ph.D. student in computer science, you should be liberally applying for graduate fellowships  even if you don’t “need” them.
What are Fellowships?
A graduate fellowship is much like a “scholarship”: a (mostly) monetary award to be used in pursuit of your graduate education.
Of course, there many *kinds* of graduate fellowships, each with their own strengths. These fellowships vary in the source of their funding, how much support they provide, their eligibility constraints, their application process, their tenure requirements, their acceptance rates, and their externalities (e.g., networking opportunities) to name a few dimensions.
Source of funding
Fellowships can come from anywhere. Most commonly, they come industry, government, and independent non-profit organizations and foundations. From industry, Google, Facebook, IBM, and Microsoft Research are just a few of the companies that provide full tuition and stipend support for at least one year. From the government, The National Science Foundation (NSF) and the American Society for Engineering Education (ASEE) both have fellowships that provide full tuition and stipend support for three years. The Siebel and Hertz foundations are both independent organizations that provide generous support to their fellows.
The source of the funding does not necessarily affect its prestige, though it might affect the externalities of the award: e.g., the non-monetary benefits, networking opportunities, and eligibility requirements of the fellowship. Some fellowships from the government might come with requirements after you graduate: e.g., that you have to work for a government lab for X years. The best ones do not have these requirements, though. Likewise, most industry fellowships are viewed, from the company providing the fellowship, as recruiting opportunities. As such, if you get an industry fellowship, you can probably expect some contact with a recruiter from that company, if not an internship offer.
Naturally, the source of the funding hugely impacts the networking opportunities that arise as a part of your being a fellow: an industry fellowship probably means you meet people (both researchers and engineers) within a company, whereas a government fellowship might mean you meet people (both researchers and managers) within a government organization.
Amount of support
Generally, the best fellowships provide full support (tuition + stipend) for one or more years. A number of fellowships fall into this category from all sources. From the government, the NSF GRFP and the NDSEG Fellowship provide full tuition and stipend support for three years. From industry, the Google, Facebook, IBM and Microsoft Research fellowships all provide full support for at least one year. Similarly, the Hertz Foundation fellowship provides full support for 5 full years.
Other fellowships may provide a fixed sum of money that can be used at your (or your adviser’s) discretion. Sometimes it’s a lot of money (e.g., the Qualcomm Innovation Fellowship provides $100,000, split between two students), other times it’s a more modest amount (e.g., the nVidia Fellowship provides $25,000).
Still other fellowships may provide a modest supplement to your existing stipend. For example, the Hertz Foundation has an option that increases its fellows existing stipends by several hundred dollars a month if they already have a fellowship from elsewhere.
All fellowships have eligibility constraints of some sort. The most common constraint is a field of study, though, since I’m primarily relaying my personal experience, basically every fellowship I have mentioned and will mention should be appropriate for Computer Science Ph.D. students.
Other requirements generally restrict potential applicants to those from specific nationalities, ethnicities, gender or stages in graduate school. Most government fellowships, for example, are strictly for U.S. Citizens. There are quite a few independent organization fellowships that were made for specific ethnicities: e.g., the Korean American Scholarship Foundation. As I said before, many industry fellowships are viewed as recruitment opportunities for the companies that provide the funding. As such, there may be very specific eligibility requirements depending on the company’s recruiting plans: e.g., only women or under-represented minorities or only students who will graduate in the next year or two.
I don’t want to convey the impression that the source of the fellowship determines its eligibility constraints, though. There are many independent organizations that only accept applications from those in their last year of graduate school (e.g., the Siebel scholarship), and there are many industry fellowships that restrict applications to only those who are U.S. Citizens and so on.
Some fellowships have very involved application processes. For example, for the Qualcomm Innovation Fellowship, I had to submit a 3-page project proposal with another student along with a letter of recommendation from a project adviser. After we were selected as finalists, we were flown out to one of Qualcomm’s Research facilities and then had to present our proposed project to a panel of judges from Qualcomm Research’s many branches in a day-long poster and presentation session. Similarly, I believe the Hertz Foundation fellowship requires a paper-application followed by a notoriously difficult interview. Other fellowship application processes are less involved. For the NDSEG fellowship, I just had to write a very brief summary of goals, provide my academic transcripts and CV, and submit a letter of recommendation from my Ph.D. adviser. I never physically saw or interacted with anybody from ASEE or the Department of Defense.
Note that a more extensive application process does not necessarily mean a more lucrative award. The QinF was a $100,000 award split 2 ways, while the NDSEG fellowship was 3 years of full support (tuition and stipend, so probably just over $200,000).
Tenure requirements are what is expected of a fellow during the tenure of their fellowship. As with application requirements, some fellowships have very strict tenure requirements while others have very lax tenure requirements. Generally, I’ve found government fellowships have relatively lax tenure requirements. For the NDSEG fellowship, all I had to do was fill out a brief survey at the end of the academic year updating them on my progress (publications, patents, presentations). In addition, I had an “external income cap” of $5,000 — basically, I could not make it excess of $5,000 from other income sources while I was a fellow. Internship pay did not count towards this external income cap as long as it was approved by the DoD.
Other fellowships have more involved tenure requirements. For the Qualcomm Innovation Fellowship, my partner and I were assigned two advisers from Qualcomm who checked up on our progress every month. In addition, we had to attend two “winners’ day” events, once at the beginning and once at the end, where we pitched our idea and presented our progress in a day-long poster session.
Generally, I would argue that having fewer tenure requirements is nicer. Having more requirements is not all bad, though. More requirements generally come with more opportunity. For example, I never got to meet any other NDSEG fellows and the fellowship did not really foster a sense of identity or community. The QinF fellowship, on the other hand, built a stronger sense of community among its fellows. The three trips I made to California as a finalist and winner were all pretty fun and I got to meet other students from all over the country.
The best fellowships are, unsurprisingly, highly competitive. I am hesitant to offer a specific number for acceptance rates, however, as acceptance rates differ greatly across sub-disciplines even for the same fellowship.
But, I can talk about a little about numbers. Broadly, there are fellowships that have open applications, those that have school-based applications, and those that have nomination-based applications.
Open application fellowships generally allow anybody who fit the eligibility criteria to apply. Most government fellowships are open application. The NSF GRFP and NDSEG Fellowship, for example, have open applications provided you are a U.S. citizen and are in your first or second year of graduate school. The acceptance rates for both fellowships are around 10% in general. For computer science, the rates may be a bit lower (I seem to remember that it was 6% for NDSEG in Computer Science the year I received the Fellowship, but I don’t recall where I found that statistic).
School-based applications are open for anybody who attends an eligible school. The Qualcomm Innovation Fellowship is one such fellowship. In its inaugural year, only students from Berkeley and Stanford could apply. The 2016 QinF, however, has expanded to include 18 schools. Acceptance rates are highly variable and depend on the number of schools allowed to apply. In the year I won, the QinF fellowship had a 6.6% overall acceptance rate (teams that won versus teams that submitted a proposal at all). I suspect this number has been decreasing over time, though, as more and more schools have been allowed to submit an application (while the number of awards has not increased by much).
Nomination-based fellowships, in my experience, typically come from industry. For example, the Google and IBM fellowships are nomination-only. Each eligible school is allowed to nominate some number of students to receive the fellowship. I’m not sure what are the overall acceptance rates for these fellowships. I can definitely say, though, that if you’re in a strong computer science school, then, it may be far tougher for you to get a nomination than get the fellowship itself. This is certainly true at CMU, and I suspect that it’s true in many other top computer science schools.
Here’s one way to think about it: it will and (should) be tough to get any one of these fellowships. But, there are so many of these fellowships out there that, provided you do a good job of writing your research statements, you have a pretty good chance of receiving some of them. If not, you may still be selected as an honorable mention or finalist which has its own benefits (which I will cover more in the next post!).
Finally, fellowships often vary in the “extras” they provide — the benefits, monetary or otherwise, apart from tuition and stipend support. For example, some fellowships provide some reimbursement for health insurance. For the 3 years I was an NDSEG fellow, I was given $1000 towards any health insurance I purchased for that year. Many other fellowships provide a stipend for buying equipment such a laptop.
Some fellowships provide non-monetary benefits such as networking opportunities, internships, or exclusive access to computing resources. For example, as I mentioned before, the QinF flies its winners out to San Diego for a day-long poster and presentation session where you get to meet not only researchers at Qualcomm but also other QinF winners from all over the country. Similarly, if you win the Facebook, Google or IBM Ph.D. fellowships, you can be almost certain that you will be on the fast-track to getting an internship with these companies. The NDSEG fellowship provided me with some access to supercomputing facilities that I never ended up using.
There aren’t too many downsides to receiving a fellowship, but I do want to mention that it’s not all just hundreds-of-thousands-of-dollars and rainbows.
When you receive a fellowship, you might be considered an “independent contractor” for tax purposes. That basically means that the fellowship will just pay you a bunch of money with no tax withholdings, so you’ll have to figure all of that out for yourself. If you’re not careful, you could owe the IRS thousands of dollars at the end of the year.
Sometimes, tenure requirements can interfere with your other plans. For example, I had to get approval for all of my summer internships by the NDSEG administrators. They never said no, but I guess they could have, and that would not have been great. Sometimes, as a Ph.D. student, it makes sense to take a non-summer internship: for example, a Spring or Fall internship, especially if you need a bit more time for your project. It would have been harder for me to do that as an NDSEG fellow because of the external income cap. Also, at the end of one of my Facebook internships, Facebook wanted to keep me on as a contractor for a few months. Because of the external income cap, however, I did not claim hours for a lot of the extra time I spent. I worked without claiming the hours, anyway, because it was important for my research, but that was a bunch of extra money that I did not get to see.
Similarly, the constant need to update my QinF advisers on our project did make scheduling time on other projects less fluid. For example, it is not uncommon for me to focus on just one project in the weeks leading up to its paper deadline. But, with an “update” meeting scheduled every month, it was harder to set aside the QinF project even facing more impending deadlines for other projects. To the credit of our QinF advisers, though, they were flexible with scheduling and did not mind postponing or even cancelling some meetings. It was just always an awkward conversation that I think could have been avoided if we were allowed to create the schedule for ourselves.
This was the first in a series of posts I hope to write about grad school life advice, in general. In the next post, I’m going to talk a bit more about why you should apply for fellowships even if you think you don’t need them. Stay tuned!
If you think I missed anything, please holla at me on Twitter (@scyrusk) or comment on this article.
One final note: If you liked this post and would like to show your support, here are two things you can do:
– Follow me on Twitter @scyrusk; and,
– Consider signing up for my mailing list below. I won’t spam, you can unsubscribe at any time, and you’ll get some content that I don’t post on the blog.
If you do either or both of those things, you’d make me happy. Thanks!
 Purposely capitalized. Academic Research is often quite different from research as it is typically practiced in industry. Perhaps this distinction is the subject of a post of its own, but the basic idea: Research is meant to be rigorous, novel, reproducible and generalizable; research is meant to (quickly) uncover key insights about specific problems and situations.
 Though you should make a solid effort at all of them by the time you graduate.
 To give you a quick example, Yahoo has awarded CMU SCS 10 million dollars over 5 years and a quick search on the NSF’s award search database suggests that there are at least 1600 computer-science related government grants worth over $1,000,000 that have been awarded to CMU faculty.
 The ones I’ve received: an NSF IGERT Doctoral Fellowship, the NDSEG Graduate Fellowship, the Qualcomm Innovation Fellowship. The ones I almost received: the NSF GRFP (honorable mention) and the Facebook fellowship (finalist).
 My monthly stipend would have been a lower, though, but not by that much.
 Really, I want to generalize this statement to graduate students in all disciplines, but I’m less familiar with the dynamics of other disciplines.