Overview

I have recently been in the spin-cycle that is the academic job search, where I’ve been applying and interviewing for tenure-track computer science faculty positions at different universities. I just finished the last interview in my batch, and wanted to reflect on the entire experience. In doing so, I hope to provide interested readers with information about what different parts of this pipeline look like, things to be aware of that might not be obvious about the process, and to share some of the advice that I got during the process. My experience on the faculty search process is for computer science, specifically in the area of programming languages and high-performance computing. These experiences and the associated advice may or may not generalize outside of this field!

Before getting started, the first piece of advice that I will give is the following: you will probably get too much advice during this process! You will have to decide which pieces of advice are worth keeping and which pieces to ignore. Especially for your job talk (discussed later), taking everyone’s advice will probably result in losing the “uniqueness” of your application and voice.

With that out of the way, let’s get into talking about the job search. I’ll break this up into several components, each corresponding to different phases of the process.

The Years Before The Search

I had a decent idea during my PhD that I wanted to try to become a faculty member after I graduated. I did a handful of things (other than the research itself) that I feel helped me a little before I actually got to the point of submitting applications.

The first is relatively obvious: getting your name out there and meeting people, so that people already know who you are and what you do before you even apply, and might already be looking for your application in their pile. Part of this is just doing good research (which is naturally hard to predict what ends up being good), but there are some more active components to consider. My advisors and I planned a small “talk circuit” across several schools I was interested in being at during the third year of my PhD, where I gave a seminar talk and had the chance to meet students and faculty. I am no doubt fortunate that my advisors had the connections to other institutions to arrange such a thing. I can’t tell exactly how impactful any of these talks were in getting an interview, but I can definitely say that in my in-person interviews, it was easier to chat with faculty that I had already met. It is surely possible to achieve this goal of “becoming known” by also networking at events like conferences (for example, my labmate Olivia is really good at this); I personally am terrible at networking and find it difficult to randomly approach people and chat.

The second opportunity I had, which I strongly recommend, was to serve on the faculty hiring committee the year before I wrote and submitted my own applications. At Stanford at least, the students on this committee had very little input on outcomes of who actually got hired, but observing the process from the inside was eye-opening. In particular, I was able to see a large number of applications and how the faculty discussed and ranked these applications. Seeing a large number of applications helped me develop some taste of my own about what makes a good or bad application, as well as set me up mentally for the quality of competition I would have on the market. Finally, being on the committee also makes you appreciate how random the process actually is — fantastic candidates get passed on for a large variety of reasons outside of the candidate’s control. For example, the department may only be hiring in a single area, or multiple areas are competing for a single slot, or the faculty already have their mind set on a different candidate, or a faculty member was not excited about a particular proposed direction, or (and the list goes on). Coming to terms with the high variance of this process was important mentally (at least for me), and strategically: if you must have a faculty position, applying widely to cover this variance is necessary.

Writing Applications

At this point, I entered the final year of my PhD, and began to write the application packet. This application consisted of two main pieces: statements (essays) and reference letters, and is usually due between the end of November and the middle of December. I’ll discuss each in turn.

Statements

A standard faculty application packet consists of 2-3 main statements. The first (and most important) is the research statement, which outlines the research you have done, and the research that you plan to do as a future faculty member. The second is a teaching statement, describing your teaching philosophy and experiences. The final statement (not always asked for or required) is a diversity statement, detailing potentially your experiences as a computer scientist from an under-represented background, and your commitment and approach to supporting diversity as a faculty member. Getting all of these statements to a relatively polished state took me around 1 month; it could have taken longer, but I got lucky with the initial structure of my research statement that significant revision was not required.

Research Statement

The most important of your statements is the research statement, which is a (roughly) three page document describing the kind of researcher you are, what research you have accomplished, and what research you plan to do if you were hired as a faculty. This is a particularly difficult piece to write, as it needs to provide value both to audiences with a variety of backgrounds, but also value to audiences who read the document with varying amounts of time.

Faculty both in and out of your area will read this document (at different levels of granularity), and should leave the reader with something to take away regardless of their familiarity with your area. What this means is that your statement should have many different levels of detail; sections discussing your prior research and overall research vision should start broad, and then narrow in with details relevant to experts in your area.

This second component of variability is the different time scales in which your statement will be read. Faculty may read only the first paragraph, or first half page, or first paragraph of every section, or other arbitrary subsets of your statement. This means that sections of your statement likely need to be organized in “front-loaded” and “sign-posted” ways, where the reader can get a feel for what kind of work you do and what kind of research you want to do even from these quick reads of the document.

The research statement needs to contain a decent chunk that describes some potential directions you might work on in the future. This part is especially important because it is what an institution is going to get if they do decide to hire you. There are a few things to think about when doing this. The first is to focus on overall research directions rather than individual projects; try to present a vision of the future, and then talk about individual projects you might undertake within those larger directions. Second, it is difficult to talk concretely about work that you haven’t done yet. This is fine! Your goal in this part of the statement is to give evidence why you are the person who can make these things happen by connecting your future vision back towards work that you have already done, to demonstrate that you have the right way of thinking or the right engineering positioning.

I’ll accumulate below some smaller pieces of advice for writing the research statement:

• Try to write an initial draft of your statement without reading too many other statements. This can make it easier to have a more unique voice in your statement, rather than sounding like other people.
• The first few sentences of your statement should try to set you up in the readers mind in the following way: “{YOUR NAME} is the person for {YOUR AREA}.”. For me, I tried to get across that “Rohan Yadav is the person for researching high-performance programming systems.”.
• If your PhD (like mine) consisted of multiple different research directions, think hard about developing a consistent story that links these projects together.
• Focus on the impact of your research. Some example questions are: What did your work enable? Do we think about problems in a new way? Do we no longer have to think about certain problems? What can we build now using your work?

My research statement is here.

Teaching Statement

The teaching statement is a two-page document detailing your experiences around teaching, your teaching philosophy (both in the classroom and as a research mentor), and the potential courses you would be able to teach at a target institution. As a PhD student, you will likely have been a teaching assistant for several courses during your PhD, as well as during your undergraduate. These are the experiences I reflected on to try to decide what were the principles that I used when teaching and interacting with students. If you have more unique experiences around teaching, such as being the primary instructor for a course, this is the place to show that off. I think the most important part of this statement is demonstrating that you have put some thought into how you teach, and how you want to teach in the future. My teaching statement is here, and is relatively dry.

Diversity / Impact Statements

The diversity statement (if requested by the school) is a one page document for you to describe your background if you came from a non-traditional background, as well as describe how you will contribute to a diversity across campus as a faculty member. I am not a diverse member of the computer science community (Indian male), and had a very traditional background and exposure into computer science. As such, I felt that the best thing for my diversity statement was to not embellish any details or promise endeavors that I didn’t have a track record of doing. Instead, I state plainly that I am committing to promoting diversity and inclusion as a faculty member, and describe with some intention about how I plan to do so. My very short diversity statement is here.

Reference Letters

Applications usually allow you to submit 3-5 letters of reference. The content of these letters is out of your direct control at the point of writing an application, but there is some strategy to consider, mostly around picking letter writers. Letter writers usually should be faculty, and the more senior and well-known the better (though these faculty should know you well before writing the letter). Some of these letters should come from your PhD advisor(s) and (if applicable) PostDoc supervisor.

However, at least one of your letters should come from someone who isn’t obligated to write you a strong letter. While on the faculty hiring committee, I remember showing my advisor, Alex Aiken, some candidates I was excited about, and asked him to evaluate their applications. The only reference letters that Alex read in this cursory scan of the applications was the letter that came from a different institution than the candidate, and wasn’t in an advisory role to the candidate. His argument for doing so was that these letter writers had no prior reason to say good things about the candidate, and was a way of getting a less biased review of the candidate, while also getting a feeling for how well-known their work is.

Given this, how do you find and build a relationship with someone like this for your own letters of reference? I think it comes back to your networking and participation in the larger research community during the years before you enter the job search. I asked Saman Amarasinghe to write a letter in this category for me. Saman is an academic grand-father (my advisor Fred Kjolstad’s advisor); we worked together for a little bit at the beginning of my PhD, but otherwise I’ve only interacted with Saman at conferences or while on a talk circuit.

My Application Spread

After putting all of the application materials together, you need to actually submit them to schools. The choice of how many schools and what schools to apply to is influenced by a large number of factors, such as where you want to live, what ranking of schools you would consider working at, what schools are hiring, or what departments you consider strong. I personally set out to apply to a relatively small set of schools (12) due to constraints I decided upon around location and ranking. This is a very small set in comparison to other colleagues I know on the job market, mostly because getting an academic job was not a “must-have” outcome for me — there are several labs I could happily do research at in industry. However, if your goal is to land an academic position with high probability, I recommend applying wider. To add to the randomness of the process, half of the schools that I wanted to apply to (6) were not hiring due to the uncertainty around academic funding and wanton attacks on academia by the Trump administration. This is a very small set of candidate schools.

Even with this small set of schools, I found it useful to make a large spreadsheet tracking the application progress for each school, tracking items like reference letter progress and which exact statements (and statement variants) each school wants. Without having a systematic way of tracking your progress, you’re likely to forget something or not properly satisfy the requirements of each school.

Video Interviews

After submitting the application packets in December, the next step is to prepare for your video interviews, which are the first round of interviews if a school likes your application. At this point, you enter the waiting game, where you (mostly) twiddle your thumbs hoping schools start to reach out to you. Some schools can start asking for interviews even before the new year, while others may reach out to you late in February. I got video interview notifications in the middle and later end of January.

A good time to prepare for these interviews is over your winter break. However, exactly how to prepare can be difficult. I prepared a set of slides to use for my video interviews and wrote down several notes about my research vision, but these ended up being not entirely useful for the video interviews that I ended up doing. What I didn’t know at the time was that every school I got a video interview from would ask for a completely different talk in the video interview!

The three video interview formats requested by each video interview I had were:

• A 15-minute presentation about my PhD research, recorded and submitted to the committee.
• A 10-minute presentation about my PhD research and future work plans, followed by a 20-minute QA with faculty.
• A 30-minute presentation, 10 minutes about my PhD research, 10 minutes about future work plans, 5 minutes about teaching and service, followed by 20 minutes of QA with faculty.

To make matters worse, the timelines between when you get the notification that you have been selected for a video interview and the video interview itself can be extremely short. I will not name names, but one school notified me that I had been selected for a video interview on a Friday night, and asked for me to do the video interview on the following Monday morning.

Making these talks might have been more difficult for me than other candidates, mostly due to the nature of my PhD research. As mentioned earlier, my PhD research consisted of a variety of loosely related topics, instead of one nice linear thread of work. Therefore, I ended up making separate talks for each video interview, selecting a different subset of the work that we (myself and my advisors) thought would be appealing to the target school (based on potential holes in their department) and the faculty joining the video interview. Because I ended up making a new talk for each interview, the slide show itself I had prepared over winter break was not useful, though the time spent organizing my thoughts was useful.

The video interviews themselves go quite quickly once you start the process. The meetings consist of you giving the requested talk, while faculty potentially interrupt with technical questions, followed by open QA with the faculty. This open QA consisted of more technical questions about the talk material itself, but also contained some more pointed, open-ended questions about research taste and future directions. Here are some example questions I got:

• How does your work differ from other people doing similar work in the field? Why are you the right person to attack these problems?
• What are some pieces of work in the community you wish you had done?
• How do you see the future of your work affected by the rise of LLMs?
• What’s a concrete project you might work on with a new student?

In-Person Interviews

After you complete your video interviews, you again enter the waiting game, hoping that you make it to the next round of in-person interviews, where you are flown out to attend a multi-day interview at the target institution. This interview consists of two main pieces: your job talk, and the various interviews with faculty and students during the day. As with before, I’ll discuss each component in turn.

Job Talk

Your job talk is one of the most important components of this interview process, as it is your chance to make an impression on a large group of students and faculty, and show off your research and future vision. This talk must be extremely polished by the time you give it at an in-person interview. I started working on my job talk after completing one or two of my video interviews, and probably spent close to three weeks of intense work before ending up with a relatively polished version. The talk itself had undergone at least 2 complete rewrites before it reached a steady state of mostly minor changes. Unlike the video interviews where making different talks for each school may make sense, I recommend making a single job talk for your entire search, due to the level of polish required. You can find my job talk here; I’m quite proud of this talk, and it even went through more revision for later interviews I did.

I’ll share a variety of general pieces of advice for making your job talk, but I’ll emphasize again the warning to not take too many pieces of advice. You want your job talk to really be unique to you, and capture your voice and style.

• Your job talk should have some appeal to a broad audience, while also giving experts some intellectual nuggets to feel that they learned something from your talk.
• You should spend around 5-7 minutes at the end of your talk discussing the future directions you want to go in.
• In the introduction of your talk, motivate the work you do by motivating your field in general, before moving into the specific problems that you try and solve.
• Try to build a cohesive and motivating story around your work, especially if your PhD work is several separate projects.
• Avoid “sounding like a student”. This can mean: referring to the larger body of work that your research lives in and focusing more on overarching problems than individual projects.
• Give your job talk to as many different people as you can, both for practice and for feedback.

For more general advice about making good talks, I strongly recommend Patrick Winston’s “How to Speak”.

Showing Up In Person

Once you show up in person, your interview will probably be two days. Each day will consist of a gauntlet of one-on-one interviews with faculty, meetings with students, lunches with faculty, a fancy dinner with faculty, and your job talk. Until you actually do one of these, it’s hard to emphasize how grueling one of these days actually is. The meetings can be intense, and the constant context switching and evaluation is stressful. At the same time, these interviews can be very fun; being able to meet so many brilliant people who have carved out pieces of time to focus entirely on hearing about your work is a very rare opportunity.

To prepare for these interviews, it is important to do research on every faculty member that you are scheduled to meet with. I collected some information about each faculty, as well as outline some discussion topics that I could have with the faculty. My advisor Fred suggested to find at least one thing I am interested in that each faculty has done; once you feel that you have something interesting to learn from this person, chatting with them is much easier.

Even though you do this preparation, what actually happens in a faculty one-on-one meeting/interview is incredibly variable. Some faculty attended your talk, and have specific technical questions they want to learn more about and argue with you about. Some faculty won’t have attended your talk and will want you to summarize your talk for them to start the discussion. Some faculty will want you to jump right in and talk about what your future research plans are. Some faculty will want to just chat and hang out. Some faculty will be completely out of your area and being able to build a bridge and talk to them is an important part of being part of the larger academic community at an institution. I remember meeting a theorist, and he started the meeting with saying, “I have never written a line of code in my life”! Preparation for these meetings is important, but be ready to let these meetings flow naturally and take the meetings where the faculty want it to go; don’t force the meeting to go along with your pre-planned discussion points. These meetings are difficult, and by the time you get to doing them, there isn’t too much more preparation that you could have done. Once you reach this stage of the interview process, you just have to be yourself!

You should have really clear and thought-out answers to questions about your future work, such as being able to discuss overall directions you are interested in, as well as concrete projects you may start working on. Additionally, practicing an “elevator-pitch” about your research will go a long way in starting off these meetings with faculty who didn’t watch your job talk.

Meetings with graduate students are your chance to get a feel of the student culture of the department and also fact-check what faculty are telling you about the department. These meetings can have an agenda, where the students have many questions to ask you, or can be free-form, where it is an open conversation. It shouldn’t need to be said, but treating the students respectfully is a requirement.

Finally, you will get taken out to dinner to a nice restaurant accompanied by some faculty. While you are still being interviewed at this dinner, it has a more laid-back feeling, where there’s more socialization and less pointed questions. Going with the flow and reading the room here is important; many of these faculty don’t get to see each other much due to their busy lives, and these dinners are a way for them to socialize as well. Dominating the discussion and turning the dinner into a lecture is a way to leave a sour taste in everyone’s mouth.

(Maybe Applicable) “A Vision Talk”

Some schools (Berkeley) have a separate talk, in addition to the standard job talk. Berkeley has a “vision talk”, which is a separate, 1-hour talk that is open to faculty only, and is supposed to be about your future research vision. This is an incredibly difficult talk to make. Making a 1-hour job talk is hard enough, and that focuses on work that you’ve already done! Making a 1-hour talk about work that you plan to do is even harder.

At this point, I don’t know if my vision talk was good, but I can describe what I ended up doing. I made a talk that outlined a larger research vision, and then identified three overarching problems in the field that I would like to see solved. From those three overarching problems, I then moved into four potential research thrusts, and identified some concrete projects within each thrust. I presented this setup to the faculty, and then dived into specific pieces that the faculty seemed interested in. I prepared a roughly 1-hour talk, and only made it through 25% of my slides due to discussions and questions raised by the faculty during the talk.

In Conclusion

The faculty job search is a long, stressful period with periods of high variability in workload and poor feedback signals around success. At the same time, it can be a fun and rewarding experience that forces you to deeply think about your research and what you want to do in the future, as well as potentially allowing you to meet many new smart people.

I just finished my last in-person interview this cycle, and am now waiting to hear back if anyone wants to give me an offer. I decided to write this up while the experience is fresh in my memory, and I hope that some more visibility into this process (and some associated advice) will be helpful for readers in the future attempting to embark on their own faculty job searches. Good luck!