I am a PhD student in Information Science at University of Colorado Boulder. I study how complex social identities, like gender and race, are represented, embedded, negotiated, and experienced within machine learning technologies — particularly, computer vision. I was also selected to be one of this year’s Microsoft Research PhD Fellows.
This academic year, I had a really successful run with PhD fellowships. I was awarded both the Microsoft Research PhD Fellowship (which I accepted) and the Google PhD Fellowship (which I could no longer accept, given you can only receive one PhD Fellowship). I was also a finalist for the Facebook PhD Fellowship. You can see my materials on my website: https://www.morgan-klaus.com/fellowships.html.
PhD fellowships are coveted awards, because they provide students with tuition and fees, a stipend to live on, and networking opportunities, all while allowing students to focus solely on their research (they do not have to TA or RA). Prior to this year, I had a not so successful run with fellowships. All of the fellowships above rejected me. (I had also previously been rejected from the NSF GRFP, when I was a Master’s student.)
So I wanted to create a post for answering some general questions about the fellowship application and interview process. I have also crowdsourced some questions from colleagues and friends that they thought would be useful for others. My advice is not necessarily fool proof; it may not apply to other fellowships outside of those offered by big tech companies (such as the NSF GRFP). I hope it simply provides some guidance in an otherwise very vague process.
What counts as a strong application?
To be perfectly honest, I have no idea. The only application I have seen is my own, and there is no feedback provided to determine what about it was strong. However, I can provide some general tips for a strong proposal that I have learned from my advisor and others through the process of writing and iterating on my application.
- Start with a core thesis. Write one paragraph which provides a focused research statement and broader research vision. In this paragraph, include what you want to do, why you want to do, and what its impact will be. Once you craft a strong thesis paragraph — make sure to share with your advisor, mentors, colleagues, etc. for feedback — you can craft the rest of the proposal around that core thesis.
- As part of that core thesis, it is useful to ensure you know what your are primarily focusing on. Is your primary focus a technology (e.g., improving the design of a specific type of technology) or a social problem (e.g., applying understanding of social issues to a technological domain)? Often, HCI research will involve multiple technical and social factors, but where your true focus is impacts the types of questions you will ask.
- Use active voice and do not be afraid to make statements that may feel bold or declarative. Passive voice and language (e.g., might, could) may come across as unconfident, but more importantly, unsure. Fellowship programs are looking to fund researchers who are confident about their research and research agenda.
- Make every sentence do work. Try to avoid using sentences which are filler sentences or reiterate previous statements without adding anything new for the reader. You have very limited space in these applications, so you should use every inch of the page to detail your research proposal.
- While you should not use filler sentences or repeat your point, you should ensure there is a clear thread throughout the proposal. A formula that I used was: (1) Start with a big broad societal issue; (2) Introduce how prior work has addressed this issue; (3) Highlight the gap in prior work and introduce your own research agenda; (4) Use the bulk of your proposal to detail this research agenda; (5) Conclude by reconnecting the outcome of your research to its impact on the broader problem introduced at the beginning of the proposal.
- Be extremely concrete when detailing your research proposal. I included broad research questions driving my overall research agenda and specific studies (and associated research questions) that would address those broader questions. For each study, I highlighted how findings would lend to understanding the broader research problem as highlighted in point 3.
- Identify recommendation letter writers strategically. One of these will obviously be your advisor, and they can help you decide who would bolster your application.
How did you frame the impact of your work?
My research is in a highly applied area — meaning, it is not something like theoretical mathematics, but has real world implications. This is likely true for most scholars in HCI and social computing. Focusing on real world implications is key, particularly when applying to fellowships awarded by technology companies (the NSF is interested in broader implications, but they aren’t the organization implementing those changes). Knowing what sort of research is conducted at those companies and what sort of products they apply is useful when framing your implications.
For example, I focused on how my research can provide useful insights in understanding and therefore improving identity representations in facial analysis technologies. I framed these implications as useful to the tech industry, including Microsoft. For Facebook’s application, I highlighted that social identity is key to Facebook as a platform, even though they do not often supply facial analysis APIs to other clients.
What you changed after your first year of applying?
If you do get rejected, do not scrap your rejected proposals, but build on them. I got a lot more focused in what I was proposing from the prior year, because I had done a lot more research in that year on my topic area. Even small language changes and building on my own work made my proposal much stronger, and showed that I had more relevant expertise to be conducting the proposed work. Even if you do not publish within the year between iterations, you should rely on your expertise, which undoubtedly will grow through reading relevant literature, conducting preliminary exams, participating in service or workshops, and working on not yet published work. For example, my MSR application from 2019/2020 seemed a bit less carefully planned when proposing actual, concrete studies I would be doing. I had a much clearer picture in the 2020/2021 cycle of what exact studies would build my research agenda for my dissertation.
What is something people do not usually know that you think is useful to know?
While you should propose something concrete, you are not married to the exact studies in your proposal if you win the fellowship. No one will be monitoring you to ensure you do the exact work you proposed, so don’t get anxiety over it.
What’s the interview process like (e.g., procedures, questions, etc.) and how did you prepare?
Of the three, the only fellowship I interviewed with was for Microsoft. Google did not have an interview process. Facebook may have, but I may not have been asked to interview as a finalist and not a selected recipient (but my feeling is they do not have an interview process).
My advisor suggested that the Microsoft interviews were very similar to meetings for faculty jobs. I had three one-on-one interviews with researchers at Microsoft Research who were within the machine learning space. The interviews were unstructured and were just focused on talking about my past work, my proposal, and my interests. Given this, there is not much to do to prepare — but it’s a great idea to be familiar with the researchers you will be meeting with.
What are the timelines for the process?
All fellowships have a different timeline, so definitely be sure to check on the ones you are interested in. I made a Google sheets to keep track of those dates each year I applied. For MSR in particular, you must first be nominated by your department. The nominations are due in June, generally. Then, if you are selected as an applicant on behalf of your university (given there are limited applicants per institution), your materials will likely be due sometime in the Fall. If you are selected as a finalist, you will then be invited to interview in the Fall or Winter. I remember finding out I had won the fellowship late December, though I was asked not to announce until the press release, which was released in the Spring.
I heard back about beginning the funding process in June. I will have to update when I know when and how the funding process actually works.