akrish

I’m going to write about fellowships before actual applications because some fellowship application deadlines are in late october and early november, which is earlier than other deadlines. The three main fellowships that people in computer science tend to apply to are NSF, Hertz and NDSEG. There are often other fellowships that are more specific to your particular research area (a slightly out-of-date list here).

I only applied to NSF, although I initially planned on applying to all three. Hertz is extremely competitive and there’s quite a bit of extra work that needs to be done to apply, such as finding a fourth recommender and writing tons of short essays. I figured that I wouldn’t get it, and it would just be a waste of time for me to apply. I don’t recommend doing this; even if you don’t get the fellowship it’s probably good preparation to think about the questions that they ask you and come up with answers to them. I think it may help a lot when writing application essays. Also, Hertz has a notoriously challenging interview and if you make it to that stage, it is probably a great experience.

I didn’t apply to NDSEG mostly because I got lazy and burnt out. The NDSEG deadline is in early January, so you don’t really have to worry about school or other applications. For me, I had just gone through my most stressful/taxing semester of college and really had no energy to write a couple more essays for the application. Also, I forgot to request transcripts, and by the time I realized, it was too late. In retrospect, I probably should have applied and I’m planning on applying again this year.

Given that I didn’t even apply to Hertz and NDSEG, I’ll just write about NSF. The application itself is just like other applications, there’s some basic personal information, transcripts, three letters of recommendation, and three essays. I may write about letters of recommendation somewhere else, but if you want some guidance in regards to them read Professor Harchol-Balter’s talk-thing. The three essays are: a description of your previous research, a research proposal, and a personal statement. The application committee evaluates your application from two perspectives: intellectual merit and broader impact. Your essays need to be written to address these two criteria.

Intellectual merit is pretty easy to address. In a nutshell, NSF is looking to see if you’re smart. If you’ve done interesting work and if you have some interesting research proposal, you’ll cover this in those two essays. Broader impact is often harder for a lot of people (and it was for me too). Since NSF is a government agency, they want to see that you are inspired to help people, the world, or at least the country. They look for things like interest in education, leadership experience, interest in including underrepresented groups in your area. They want your work to advance society or otherwise contribute to
scientific knowledge.

It can be hard to address broader impact in your essays. One “easy” way to do it is to actually do some stuff as an undergrad that you can talk about. For example, you can be a TA, or better, you can volunteer at local middle or high schools. I’ve heard about undergrads that host these robotics programs at local schools in Berkeley, which I think is a really great thing to do and a great thing to write about. I didn’t really have anything like that to write about, so I tried to spin my research goals as providing an infrastructure for future research. For me, addressing broader impact was definitely the hardest part of
the application. I spent a lot of time figuring out how I would write my personal statement so that I could (in my opinion) cover it. For you, it may be really easy. Since the previous research statement and the research proposal are meant to be quite technical, generally people write about broader impact in the personal statement.

THE ESSAYS

The previous research statement was fairly straightforward for me. I had already prepared an outline of the previous work I done to help my recommenders, so I just transformed this outline into an essay. Since I’ve worked in both security and compbio, I had a bit of trouble trying to unify the essay and explain why I chose compbio over security for the future. My guess is that this probably won’t happen to you. My thoughts on this essay are:

1. Be technical. They want to see that you can communicate your ideas effectively (b/c presumable you’re going to be a leader in your field) and in particular that you can communicate technical ideas effectively. It’s often hard to talk about your work when you abstract away the details, and this may negatively impact the clarity of your writing. At the same time, don’t drown the reader in the details; make sure the high level ideas of your work are clear. For example, I described the technical formulation of the problem that I worked on, but only presented the high-level ideas behind the algorithm we developed. I guess there’s a balance here, but the most important thing is to make sure your ideas get explained clearly; if you have to be technical to do it, then that’s probably fine.

2. Talk about work that you did. If you worked on a project with a graduate student or as part of a larger group, write a bit about the project as a whole, but then talk about the specific problems that you solved. NSF is going to fund you, so they want to see that you can do great work. They don’t really care if your group (or your grad student) can do great work. This can be slightly challenging because you probably aren’t working on super challenging or revolutionary problems by yourself as an undergrad. However, I made small contributions to a larger project that I thought were interesting research and I wrote about them.

The research proposal can be a really hard essay to write. NSF wants to see that you can come up with an interesting research problem and propose a reasonable solution for it. I think it’s really hard to figure out what interesting problems are (and so do a lot of people), so I recommend that you talk to a lot of people about this. Ask your advisor about what’s going on in your area, talk to graduate students other faculty etc. Find a problem that is well-motivated and hopefully related to your prior work. At the same time, you probably shouldn’t write about something revolutionary that will change the world. For example, don’t propose to solve P vs. NP (unless you already know how), people just won’t take you seriously. I chose a really small problem that came up in my compbio work. I talked about that work and said that this was a problem with our existing approach, then I proposed a completely different approach that seemed reasonable. I was fairly technical both in describing the problem and in writing about my approach. I also tied it into some broader impact by saying how the work could benefit society.

The personal statement was definitely the hardest essay for me to write. A large part of this was figuring out how to address broader impact. I wrote about a bunch of random stuff like being a captain on my ultimate frisbee team, teaching a databases class, building a compiler as a class project, and of course, my research. These were loosely related by some learning and adapting that was going on each setting. I brought in broader impact by talking about how computational biology should be a platform for biological research, and that I’m interested in building the robust tools that make this platform possible (an infrastructure for future research). I also talked about how I enjoy teaching and how I think my specific interest (which was finding genes associated with specific genetic diseases) could play a small part in understanding the biology behind genetic disorders. This essay was definitely less technical than the other ones, although I did get into some specifics about my teaching, class projects and research.

The essays are basically the meat of the application. I think it’s worth spending a lot of time on them, because it’ll give you a wealth of ideas that you can tie into your other application essays. My actual personal statement (for schools) was kind of just an amalgam of the three NSF essays I wrote and writing them ahead of time meant that I had all the ideas I wanted to talk about already and just needed to tie them together.

Another large part of the application is the letters of recommendation but I’ll write a bit about that later.

Also, when I get a chance I’ll upload my essays here.

For those of you who don’t know, I just finished applying to grad schools and am currently in the process of deciding where I want to spend the next 4-n years of my like (where n >=4). I’m applying for a Ph. D. in computer science and my research interests are in Machine Learning and Computational Biology. During the whole process I got a lot of help from Philip Guo and Jean Yang, so I thought I would do a brain dump of my thoughts on the subject.

Naturally, a lot of this will be about me (for example this post will be about why I decided to go to grad school) but maybe some of that will be useful to you. To get some other data points about it, read what Philip and Jean have to say. Ok here we go…

In a nutshell, I’m going to grad school because I like to do research. I figured out that I like doing research by, well, doing research. As an undergrad I’ve been researching for a couple of years with David Wagner and I also got to spend some time working with Ron Shamir. Before starting to work with Professor Wagner, I really didn’t think I’d be interested in research, and honestly, I didn’t really know much about the academic environment and research at all. I was really excited about entrepreneurship and wanted to start a company, but one of my goals in college was to try a lot of different things, so I figured I’d give research a shot.

Another reason why I got involved in research in my sophomore year was that I didn’t like the internship I did the summer before. I worked for a fairly small company (~300 people) but I felt like I wasn’t doing anything important and I wasn’t really doing work that I found interesting. After my internship, I thought that maybe industry wasn’t for me, and the other natural thing to try was research.

So I got involved in research and discovered that I liked it for several reasons. (1) I get to work on really cool problems, (2) I get to think very hard about all sorts of possible solutions, and (3) I have the time to find and work with the best possible ones. Also, (4) I get to be (fairly) independent and (mostly) in charge of my work, and (5) I (for the most part) don’t have to do things that I don’t feel are important. In my internships, a lot of these factors weren’t there, although I do think that this may have been a product of my specific internships. I know people who do really exciting work at all sorts of companies, I just personally haven’t experienced that.

Given that I like research, the best place to go is graduate school. So by the end of my junior year, I knew that was what I wanted to do. I somewhat arbitrarily decided that I wanted to research in compbio and machine learning. A big factor in this decision was my summer research in computational biology. I got to design and implement really cool algorithms and I realized/discovered that there were tons of super-interesting problems in biology that could only be solved by computational methods. I got to read a ton of papers about these problems and found that often times there were elegant solutions to them. Moreover, I was part of a great group that involved me in their projects and told me about the awesome work they were doing. All of these things influenced my decision to go to grad school and helped me decide on computational biology as a research field.

I applied to many (9) schools. This is in many cases more than a lot of other people I’ve talked to. The two reasons for this were that (a) I didn’t have any existing job offers and (b) I didn’t think I would be happy in industry. I personally would have preferred to go to top-20 grad school over getting a job in industry, and this is reflected in the number of schools that I applied to. I know a lot of people who only apply to 2-4 schools and also look for real jobs.

And that’s basically it. After deciding to apply and where I would apply, the next steps were to get my applications together, write some essays, and wait to hear back.