Applying to be a Professor’s Apprentice
Hi, I’m Pranav- a soon-to-be-graduating Dual Degree student in the EE Department and an incoming Ph.D. student in the Human-Computer Interaction Institute at CMU. This is an aggregation of my thoughts on applying to grad school and resources that I relied on during the process. There’s already a lot of fantastic advice out there by people far more informed than me, so to avoid redundancy, I will link to those resources wherever possible. My main aim here is to unpack some of that advice for international applicants, share my experiences, and talk about things that worked for me in the process. A lot of this advice is specific to Ph.D. programs in HCI/Computer Science/Information Science and I’d caution against generalizing it to other programs and departments. As an additional disclaimer, I have obviously not served on the admissions committee anywhere so this post is largely based on conversations with people who’ve been on those committees, my own experience applying, and other popular advice out there.
Before I start, Kalpesh Krishna has curated a lot of great advice/online resources here and you should definitely check it out. It contains a lot of application guides and admission notes for different CS Grad programs. A bunch of people also came together to write this comprehensive guide on applying to NLP Ph.D. programs and is a great resource if you want to get an idea of what the process looks like currently. Philip Guo has written a comprehensive guide on applying to Ph.D. programs here and Jean Yang has written about her experience and the process here. I referred to a lot of these resources myself and will link to them throughout this post as well.
Deciding to Apply
Most people on the fence about applying to grad school usually have similar reservations. The most common reservations are whether grad school might be a good option for them, whether they should choose to go to grad school over the other options they have at the time, whether they should wait for a while and gain some industry experience before applying, and lastly, whether they are actually “ready” to apply and might get in if they did.
To grad school or not to grad school?
Should I apply to a Ph.D. program? Should I consider getting a job? I’m good at Krav Maga, should I go pro?
— Not Pranav Khadpe
A good starting point would be to assess whether a Ph.D. might be a good idea, independent of what other options are in the loop. This is easy to evaluate as a senior undergraduate because I assume everyone has had some research experience by the time they turn in their thesis and it should be fairly straightforward to assess whether or not you enjoyed the process. I had done two research internships by the time I was applying and had wrapped up two big-ish projects and another smaller one. I had enjoyed working on all of them and more importantly, I enjoyed the freedom afforded to me in terms of being able to take initiative, lead a project, and come up with and execute solutions independently. My positive experiences directly influenced my decision to apply. Your experience, unfortunately, is strongly linked to who you worked with, their style of working, the problems you worked on, and the success you’ve had doing research. Most of the people I know that claim to “enjoy” research have had some sort of success in getting papers accepted or securing some competitive award and it wasn’t any different in my case. Though it’s hard for me to convince you to keep pursuing research if you haven’t had any source of gratification, I’d still encourage you to consider grad school because it’s possible you’ve just had terrible luck. If you’ve had a sub-par experience doing research so far, haven’t had sufficient research experience to take a call or are a junior student interested in getting into research, I’d recommend checking out Philip Guo’s talk — “Why Pursue a Ph.D.?”
He points out three big incentives- freedom, making a name for yourself, and more jobs to choose from. If any of those reasons resonate with you, then you should probably consider applying to grad school. Jean Yang has also written about it extensively here. You could also read accounts of what pursuing a Ph.D. is actually like here and here to see if you might enjoy that kind of life. Ultimately though, you’ d only be able to take a call after having done some research (also a prerequisite to getting accepted to programs) and deciding whether or not you want to do it full-time. So seek out opportunities to do research and start exploring early. In my opinion, it also doesn’t matter what area you start out in. I spent second year working on analog circuits and VLSI and third year working on language technologies before settling on HCI. The research process is largely similar across areas and as one of my mentors points out- a Ph.D. is a Doctor of Philosophy in something. You’re learning how to ask good questions and do research and then extend those methods to the something. So don’t stress too much about the research area when seeking opportunities- if it’s exciting, get involved.
How to evaluate doing a Ph.D. against the other options you have is a hard question to answer. It is similarly hard to exhaustively explore all your options and pick the one that’s “best” for you. We’re all spoilt for choice and understandably so- a lot of us chose to study engineering because of all the doors it would open up. Jean Yang’s advice on how to navigate this “quarter-life crisis” probably helped me the most. The thesis of her advice is that you’re better off picking something and committing to it. Pick whatever you feel most excited about at the time and go all out. If at a later point, you realize that you are not as interested as you were initially, you can switch to something else. And it’s easy to make that switch if you’ve proved yourself once before. You’ll learn as much, if not more, about whether you like something by actually doing it, rather than thinking about it. Like others before me, I felt 22 was an appropriate age to go explore and decided I’d commit to exploring whether full-time research was a good idea for me. Though I do still maintain that committing to grad school might be the least risky of options to commit to. That is primarily because if you realize that a Ph.D. isn’t for you after you’ve started, you can still drop out and go get that software engineering job but it’s harder to go the other way around. So, though there is obvious merit in working for a while and getting a better sense of your interests, it might make it harder to get into grad school. Especially, if you’re an international student. In a sense, committing to a Ph.D. is known territory. You’d be doing what you’ve always done- keeping more options open.
Jean Yang has spelled out her own decision-making process here.
How did I make my decision?
I’m going to stress that deciding to apply is an entirely personal choice. There is no right answer to the “why”. It could be because of a bet you made. It could be because it sounds cool. It could be because you wanted to get a job that pays a lot, after. It was not my general affinity for 3 letter acronyms that drove me to apply for Ph.D. programs. In addition to my interest in pursuing research, I think I applied for a Ph.D. mainly because many people I admired and hoped to emulate had Ph.D.s. There is also a ceiling on career growth as a typical RnD engineer and so it made sense to get a Ph.D. looking at it through the lens of career-optimization. That being said, your success in a Ph.D. program will be a function of your motivation and if the motivation is intrinsic you are more likely to succeed (So I’ve been told. I wouldn’t know — — I haven’t done a Ph.D. yet). Programs admitting you are also making a bet on you succeeding so they are usually trying to figure out who’s intrinsically motivated and who isn’t while evaluating your application. If you’re keen on doing a Ph.D. because you want to make a lot of money or want a high paying job after, what happens if you’re made an offer with similarly high compensation during? Do you drop out? What happens if you win a lottery? Ideally, you’d want your motivation to be minimally dependent on extrinsic factors.
Are you ready?
Applying to grad school is definitely a cost-prohibitive process (each application will cost you roughly a hundred dollars and the GRE and TOEFL are another couple hundred dollars) and it makes perfect sense that you’d want to take the dive only if you feel “ready” and feel confident about getting in. Generally, a program will only admit you if they think you’re ready so I’ll talk about your chances of getting in as a proxy for “readiness”. (*drumrolls*) It’s hard to predict whether or not you’ll get into a Ph.D. program because there are way too many variables at play. Each school has different selection criteria and there’s an obvious element of luck (whether a certain professor is recruiting that year, who else is applying, whether an ongoing pandemic has led to reduced department funding). Based on my experience and the advice I got, two factors far outweigh the others:
- Demonstrated research experience: If you’ve done research it greatly improves your chances but notice that I said “demonstrated research experience” and not just “research experience”. Whatever you claim you’ve done needs to be evidenced by published papers/drafts/open source software/artifacts or by credible references talking about it in their letters. The admission committee has no reason to believe what you say and unless there’s evidence, it counts for little. Most people I know, that applied an got into Ph.D. programs in HCI had at least one top tier paper.
- Who you know and who they know: Most universities are risk-averse and will always prefer an applicant backed by academics they know. That puts international students at a disadvantage and is something you should be conscious of. A letter from someone known to the person reading, it is easier to calibrate and so your chances improve a lot if you’ve worked with and have a letter from someone like that. As Mark Corner puts it https://emeryberger.com/admission-notes/: “There are a few academics at the Indian IITs that are well known in some fields. This means that although letters from your college are required, they may not be very helpful if the person who wrote it is not known in the US. It is very well known in the USA that foreign students often write their own letters and have professors sign them. This makes them mostly useless unless that person is internationally known and trusted to write their own letters.” [sic] He even mentions that a letter from a US academic carries more weight than your research, SOP, undergrad institution, and grades. Which although blunt and discouraging, is true.
Your grades, GRE, and TOEFL matter but not as much as you think. Most places no longer require the GRE and those that do use it as a filtering criterion. So while you need a good score (In my opinion anything over a 325 is all the same), an astronomical score probably won’t help. TOEFL scores are mandatory for international applicants but again, they are mostly used to filter out obviously bad candidates (The highest cutoff I’ve heard of is a 110 with a minimum of 26 in Speaking). And of course, don’t shy away from applying because of a low GPA. It’s possible to offset that with experience and strong letters. I’d generally recommend that you keep the CGPA above 9 to be a competitive applicant at some of the more selective programs but at the same time, I know of people with late 7 pointers making it to Cornell, CMU, and MIT.
Additionally, some people express concern about their research interests being too broad which in my opinion shouldn’t be a deterrent. Now I’m not suggesting that you should dive into a Ph.D. if you’re interested in “Robotics” & “Computer Vision” & “NLP” but if you’re interested in “Multilingual NLP” & “Conversational-Systems” & “Computational-Sociolinguistics” that’s probably fine. Unlike Ph.D. programs in Europe where you’re expected to come in with a razor-sharp focus, Ph.D. programs in the US expect you to spend the initial years navigating your interests until you settle down on what becomes your dissertation. While “stating your purpose”, you should be able to chart out a loose plan of what you intend to do for the next 5/6 years and not necessarily your entire life’s research mission. At the same time, you need to demonstrate that with some supervision you might succeed at executing said plan. That’s it. So pick one or two things you’re interested in and center the SOP around them. Once you get in nobody is holding you to your SOP, so it’s fine if there are still things you want to explore. Everyone understands that research as an undergrad is opportunistic and you probably had little say in deciding the problems you worked on. It’s unreasonable to expect you to have clarity on your research plans and figuring out what your interests are is part of what the Ph.D. is.
To get a better idea of how applications are reviewed at different places, you could check out :
- Philip Guo’s five-minute guide
- Mark Corner’s Notes on UMass Amherst admissions
- Jason Eisner’s Guide on JHU CS
- Scott Fahlman on CMU LTIs admission process
A research-based MS at a US university appears to be an obvious workaround to most of these issues as it would help you gain research experience, network with faculty, and help you concretize your interests. Using the MS as a gateway to a more selective Ph.D. program is also not uncommon. An MS, however, does not usually come with the funding guarantee that a Ph.D. does. While it is pretty common for international students to get a TAship/RAship after starting, there is an obvious element of uncertainty. In my case, I didn’t consider applying for an MS. I had a letter writer who was well known in the field and I had one published paper as third author and two under review- one as first and the other as second author (Two were at top tier venues and all three were a part of my application file). I felt confident I would get in somewhere and so decided against applying to MS programs. If you are considering applying to MS programs then you should probably single out programs where it’s easy to secure funding. I know UIUC’s MS CS is pretty popular for being generous with funding. CMU also has some research masters programs like the MS in Robotics and the MLT where funding is usually easy to get, though not guaranteed. You should also keep in mind that MS applicants are evaluated differently and you can probably get a better idea of that here:
- Karthik Raghunathan’s guide on Stanford’s MS CS admission process
- Shriram Krishnamurthi’s article on financial aid in MS CS programs.
Finally, even people who are ready to apply to Ph.D. programs and might be strong applicants tend to second guess themselves. I did. Most of my friends applying with me did. It’s a great time to question your potential, capabilities, and for self doubt in general. My mentors, friends, and family did a good job trying to reassure me but if you need to hear it from someone else here is some well-articulated advice you should definitely go through:
The Application Process
If you’ve decided you want to apply to grad school then you’re looking at the long process of applying which is extremely draining but also rewarding. In this section, I’ll go over the process of getting your application together and the process after applying.
Taking the GRE and the TOEFL
Since these two tests are essential and are unrelated to your interests or schools, I’d recommend getting these out of the way as soon as possible. There’s no prerequisite and so you can take them whenever. I’d recommend taking them before the end of the Spring semester of your pre-final year. I took my GRE in Feb 2019 and TOEFL in Aug 2019. For reference, my applications were due Dec 2019. I spent three weeks on GRE prep alongside the usual academic load and managed a 328/340 (Verbal:161, Quant:167). The TOEFL is much easier and you’re fine just “showing up” if your English isn’t too shabby. I had a 115/120 (Reading:30, Listening:29, Writing:29, Speaking:27) on the TOEFL. You don’t want to spend too much time on these tests because like I’ve mentioned the scores don’t matter a lot.
In my case, most of my awareness of relevant programs came through reading papers in my areas of interest and looking at the affiliations of authors whose work I liked. I also benefitted a lot from talking to current Ph.D. students at different places as well as talking to other people applying with me. You could also talk to your research mentors/advisors and ask them to point you to programs that might be good for you. The main takeaway here is TALK TO PEOPLE. That’s the only way to improve your awareness. Don’t shy away from talking to people and talk to many people. Another valuable resource is csrankings.org. I used it to shortlist professors at different universities whose interests aligned with mine.
You also want to try and find out which faculty are actively recruiting your year. This is a bit hard to find out but a good place to look is twitter, where faculty often advertise if they’re recruiting students. You can also try finding out from that particular professor’s current students. In the interest of time and money, you don’t want to end up applying to a school to work with just Professor X when Professor X doesn’t have the bandwidth to take more students. You might also consider writing to faculty you’re interested in working with to talk about what they’re currently working on and what their group’s current focus is, although faculty are usually unlikely to entertain such correspondence. Some professors will explicitly mention on their page if they want to hear from prospective students and you should definitely reach out to them if you want to talk. Be sure to make it worth their time and remember that you have to make a good impression so do your research. Another thing that worked for me was asking people I’d worked with to introduce me to faculty they were in touch with. This definitely improved my chances of hopping onto a call with them.
Once you’ve found out about relevant programs, you should apply to places you’d be happy attending — — you don’t want to apply to more than 10 places because that would be a lot of work and really expensive. What factors drive that decision on where to apply are entirely up to you. What will most likely happen is you will be able to weed out places you definitely don’t want to attend, have a list of places you would definitely attend, and you might be on the fence about others. For places that you’re on the fence about, I would recommend applying to those places anyway and deferring the decision of attending to after decisions are out. If you want to apply more conservatively, make sure a few of the programs you’re applying to are “safe”- where you should be able to get in no sweat. A good way to figure out which places might be safe for you is to talk to your advisors/letter writers and ask them where you stand a good chance of getting in. Additionally, while applying to a Ph.D. you’re not just applying to the university or department as a whole but to a certain faculty at that university and so if a faculty is cross-appointed across departments, apply to the one that is their primary affiliation (eliminates the need for a co-advisor) or the department that you most associate with(will determine your cohort).
I ended up applying to 9 Ph.D. programs.
Talk to your letter writers
Talk to your letter writers early. Give them a heads-up on the programs you plan on applying to so that they can budget their time accordingly. It’s also important to talk to them about your research plans and what you’d like them to highlight about your work with them. You also have to be careful about who you ask to write your letter. Philip Guo has articulated it well under “Who should I get to write my letters of recommendation” here. Jean Yang has advice here. I spoke to my letter writers around mid-October, shortly after finalizing the programs I planned to apply to. I had the same 3 people write me letters to all the programs I applied to- my advisors from each of the major projects I’d worked on. I also know people that requested different people to write them letters depending on the program they were applying to. Sometimes, professors might make a fuss about writing a letter to more than n places and so you might have to rope in someone else for the remaining applications. Strategize accordingly.
Writing your SOP
Writing a good SOP takes time and you should start as early as possible, get as much feedback as possible, and keep iterating on it. Some people advised me against reading others’ SOPs as it would “bias” me and while these people have been successful applicants themselves and their advice has merit, I decided I’d get rid of the “bias” by reading a LOT of SOPs. This is because while each SOP is different, there is an underlying structure that seems to hold across them and admission committees have an expectation of what they should look like structurally. Here are some resources I referred to on what a good SOP should look like that also contain sample SOPs of a lot of successful applicants:
- Christopher Fletcher’s Guide on Personal Statements
- Philip Guo’s SOP and some of his students’ SOPs
- Jean Yang’s SOP and her advice
- Ethan Fast’s SOP and his advice on writing a good one
Here are links to my own SOPs for CMU and Harvard (where I managed to land an interview but ultimately got rejected). I’m linking two of them so that you can see how much or how little changes depending on the school. Some schools want bespoke material and will have different guidelines on writing the SOP. For instance, CMU required that I mention my areas of interest in the first sentence but it’s usually incremental effort to apply to one more place after you have the main content ready.
I enjoyed writing my SOP because it was a good forcing function for me to sit and think about the kind of problems I was excited about, reflect on what I’d worked on so far and figure out what direction I wanted to take my research in. As an undergrad I often found myself rushing to find “the next thing” to work on without slowing down to think of what my research agenda is or what my real research goal is and I think writing an SOP definitely helped me get a better sense of my own goals.
Personal History/ Diversity Statement
Some places like Berkeley, UMich and UW ask you to write a Personal History Statement or a Diversity Statement and this is a good place to talk about any hardships you’ve had to overcome in accessing education, what diverse perspectives you bring to the program, how your background has uniquely shaped your interests. Since I recognize myself as coming from a fair amount of privilege, I focused my statement around how I ended up doing the research I’m doing and why I care about it. My SOP, on the other hand, focused on concrete research experience and what I’ve done so far. There’s a minute difference between the two but make sure your personal statement isn’t just another SOP. Here are some sample Personal History Statements.
Deadlines are typically between early to mid-December and what follows is an agonizing wait till around mid-Jan when you start hearing back about interviews. It is the norm at almost all places, for professors(usually the one you’ve indicated interest in working with) to interview shortlisted candidates as the final step in the selection process (CMU HCII doesn’t). I had interviews at 6 of the 9 programs I applied to and from what I’ve heard places usually interview between 2 to 3 students for every position they want to fill.
Most of the interviews were casual conversations and faculty were interested in getting to know more about me and why I wanted to pursue a Ph.D. while also assessing if I’d be a good fit for their group. The only technical questions I was asked were about the work I’d done but these were few and not unexpected. They were mostly looking to see how closely my research agenda matched with theirs and whether we would be happy working with each other.
Interviews are also a great time to network with faculty and for them to get to know you so don’t be too nervous about them. Try to make the most of the uninterrupted time you get with them. You could get feedback on your work, discuss ideas you’ve been thinking about, and ask them questions about their work. It’s important to remember that while you may join a particular school to pursue your Ph.D., you are signing up to be an apprentice of the field as a whole (HCI in my case). So, it’s important that you network with faculty and stay in touch with them regardless of whether you end up joining their group or not.
Decisions and Wrapping Up
After the interviews, you are propelled into another 3 week-long wait till decisions come out. Most of my decisions arrived by mid-Feb and I ended up getting into 3 of the programs I applied to and decided to commit to CMU. Though my friends probably saw me at my worst behavior between mid-Jan and mid-Feb when I was extremely stressed out about the rejections that had started coming in, I was pretty happy with the way things turned out in the end. I’m extremely grateful to them for tolerating me when they didn’t have to- I can’t thank them enough! I’d also like to reiterate that there is a lot of randomness in the process and it’s important that you don’t get too disheartened by the outcomes. I also acknowledge that it’s easy to give such advice but very hard to follow. I was pretty gutted by a couple rejections and anyone passing by my room could probably tell from the very audible expletives.
I’m doing a disservice by not talking about my decision-making process in detail and that’s partly because most visit days- usually in March- were canceled this year; it has been a pretty atypical year. You will find a better account of the decision-making process here and here. Andrew Kuznetsov has also curated a great guide on picking a Ph.D. advisor here. One thing that attracted me to CMU was it’s “free agent system” where you don’t commit to an advisor going in but align with one once you’re there. They’re also pretty accommodating when it comes to switching advisors. Some people think of it as non-ideal and would prefer knowing beforehand who they are going to work with but given my broader interests and desire to explore a bunch of ideas, it made sense for me to join a program that offered me the flexibility.
If you’ve made it this far, thanks for taking the time to read this, and I wish you the best of luck with your applications! If you need to reach me, you can write to me at email@example.com or find me elsewhere on my online estate:
The latest Tweets from Pranav Khadpe (@PranavKhadpe). Incoming PhD student at @cmuhcii | Electrical Engineering…