The academic job search for computer scientists in 10 questions

Nicolas Papernot and Elissa M. Redmiles

"Piled Higher and Deeper" by Jorge Cham

www.phdcomics.com

Are you about to wrap up your dissertation or finish your postdoc appointment? First of all, take the time to appreciate all you have accomplished so far; this will help you make the most of the upcoming months. The (academic) job search can be an exhausting process but it’s also a unique and exciting opportunity to build a broader scientific community and present your research to this new network of researchers. In this post, we focus on the academic job market for computer science, but some of the advice on how to prepare your application materials or interviews may apply to other fields as well.

Resources contained in this post

Why should I apply to academia? There are so many great opportunities in industry.

If you are graduating with a PhD in computer science, you might wonder how opportunities in academia and industry compare. In the end, this boils down to personal preferences and depends a lot on your research interests. If you are considering industry primarily for its research labs, I would however recommend taking the time to explore both options (academia and industry) completely, even if you feel like you are leaning towards industry. One of the reasons is that interviewing for tenure-track positions in academia will expose you to a lot of people in the community---especially researchers that you don’t normally meet at conferences because they work in areas further away from your research interests. Typically, you will meet with about ten to twenty faculty at each university you visit. This will not only be a great way to get feedback on your research from a very different perspective, but also to learn more about the cutting edge problems that other researchers are working to solve. For instance, I learned about some of the challenges faced by the graphics community during my job search. Of course, if you are attracted by industry because you value having an impact on the product and customers rather than publishing your research, you should focus your energy on the industry job market (and can probably stop reading this blog post unless you’re curious or in the mood to procrastinate).

Alright you convinced me. What is the timeline?

Applying for a tenure-track position in academia does require a bit of preparation. It’s good to start thinking about your job search a year before your expected start date. This will give you enough time to prepare your job package (typically between August and October), reach out to your letter writers and most importantly seek feedback on your application material.

This brings me to the timeline. Applications to tenure-track positions are usually submitted between November and December. You will start to hear back from universities between December and February. At this point, they will either invite you for a phone interview or directly invite you for an on-site interview. This typically depends on the procedure of the university, not your strength as a candidate. Next, you will interview on-site during the months of February, March and April (this timeline has been expanding in recent years, to include interviews in January and May depending on the market and institution). As offers start coming in, you will also start making decisions in April or May.

Instead, the process in industry is much faster: you will generally interview within a week or two of applying and hear back within a week or so from your on-site interview. People move fast in industry, and they might feel odd if you postpone interviews by several weeks. That is why I recommend applying to industry only when you are about to finish your academic interviews. This will allow you to receive offers from both academia and industry roughly at the same time and make an informed decision between the two of them.

The one caveat to this is if you wish to take a one-year postdoc in industry before continuing on to an academic position. This has become an increasingly popular trend in recent years. Taking this one year “pre” sabbatical gives you exposure to real-world problems in industry, a year to do research without funding and reaching pressures, and time to recruit students before you start your position. However, doing so also may require moving twice in a short period of time and extends your time until tenure.

If you decide that a one-year postdoc in industry is a good idea for you, you may want to interview for the postdoc while you’re on the academic market, so that you can tell universities where you receive an offer that you would like to defer for a year and where you’ll be spending that year. These postdoc interviews may be quite formal (e.g., at MSR the postdoc interview is 1.5 days on-site) or very informal (a phone conversation with your future postdoc mentor).

Where do I apply?

Coming up with a list of universities to apply can be difficult and the list will vary from person to person. Typical considerations include “two-body problems” (we’ll refer to this as “dual career” in the remainder of the post, since calling someone’s spouse a problem feels a bit rude) and location preferences. Some people are open about dual career considerations in their job search - ranging from noting this situation in their cover letters to mentioning it during the interview process - while others wait until they receive an offer to even mention that they have a spouse. Similarly, some people apply only in locations where they know they would accept a position, while others apply broadly and narrow down at the interview or offer stage.

A good resource for finding open positions is the CRA job posting, which you can subscribe to in order to receive a daily or weekly digest if available academic jobs. You can sometimes find this out from the postings advertised by each university’s recruiting committee. Your advisor(s) and senior researchers in your community are also good points of contacts to establish the list of schools that would be good matches with your profile.

This is a good time to start a spreadsheet that tracks all schools to which you apply. As you browse each university’s job postings, be sure to write down all the material required as part of the application package (e.g., research statement, diversity statement, …) in addition to the application deadline. This varies from school to school and it is very useful to have a snapshot of all of this information in hand to make sure you prepare the right material and that you don’t miss the different deadlines. In particular, the list of universities you will apply to should be ready by mid October at the latest to ensure the earliest deadlines can be met.

Let’s get started. How do I prepare the job package?

Most places will ask for a subset of the following information and documents: cover letter, research statement, teaching statement, diversity or broader impact statement, a list of letter writers, sample publications and/or list of publications.

I decided to make my cover letter very short because it felt like otherwise I would repeat a lot of what was already included in the other statements. My letter made it clear who I am in a few words. The research statement is a difficult writing exercise: it needs to reflect your level of technical understanding of the research topic while at the same remain accessible because it will be read by faculty outside your research area. 

Nicolas: I felt it helpful to structure it around three sections, each describing one direction of my research. Each section started with a highlight of the work: a few sentences accessible to anyone in computer science describing the research direction at a very high level. I wrapped up each section by stressing how the work presented impacted the community (e.g., with open-source code). The teaching and diversity statements are more personal and it is hard to give generic feedback on how to write them.

Elissa: I structured my statement a bit differently. I started with an overview of my work, including a clear paragraph on broader industry and policy impacts. I then discussed epistemological work I had conducted up front: the validity of work with human-based data is often questioned, so I wanted to deal with those issues immediately by describing the work I had done to examine the validity of human-based data in security. I worked to tell a story with the remainder of my thesis, tying in even my non-security work to one main story line. The exercise of creating this story was very difficult, but extremely helpful for preparing my job talk and interviews, where I needed to articulate a clear story connecting what could look like a disparate set of research themes. You can find a copy of Elissa’s research statement here.

Your choice of letter writers is important. They need to know your work well enough to describe it and comment on the quality of the work. You may want to pick different types of letter writers, to the extent that you can: people at different levels of seniority (although the more recognizable the person, the better), at different institutions, in different CS sub-fields, and, if your work is interdisciplinary, from different fields. I recommend having at least 5 references so that you can spread the load among them: they often have to customize letters for each school that you apply to. This and the fact that each school tends to have its own way of collecting letters, your references will spend a lot of time to support your application.

You may want to create a letter writer cheat-sheet to help your letter writers understand how they fit into the landscape of your application: Who are the other letter writers, what are you hoping this letter writer will comment on in particular, what are you emphasizing as the “story” of your application? Be sure to keep track of who you included as your letter writer in each application;different schools will ask for different numbers of letter writers. You may also want to create a tracking spreadsheet for each letter writer (credit to Eszter Hargittai for the spreadsheet example & idea) to track the letters they need to submit for you and where they should submit them. This is especially helpful if you have letter writers from different fields / backgrounds who may be familiar with different application systems than those used by the institutions to which you are applying.

Don’t forget your website. A few professors confessed to me that they only had a chance to peak at my website for a few seconds before I entered the room for our one-on-one meeting. Having an accessible website highlighting your recent work, outreach and service to the community is definitely helpful throughout the job search. To the extent that you can prepare this website well ahead of the job market, this will help you be easily visible to universities that are potentially interested in hiring you.

What is the goal of the job talk?

The job talk is the #1 most important part of your job application. While your materials (aside from the letters and maybe the research statement) will likely be quickly skimmed over, nearly all of the faculty in the department will attend your job talk. You will have their undivided attention for one hour, during which time you need to convince them that they want you as a colleague and that your work (a) is yours, (b) is scientifically robust, (c) is interesting, (d) that you know what to do next, and (e) that you can handle questions comfortably and with appropriate decorum. Let’s break those five points down:

  1. Showing that your work is yours: A primary question coming into a candidate’s job talk is whether they were driving the work or whether their advisor -- or some other coauthor -- was driving the work. To address this question, it is best to focus your job talk on first-authored papers, include any single-author papers if you have them, and be prepared to calmly and comfortably answer questions about your research (more detail below). You can also choose to add references to your papers at the bottom of your slides so the audience understands how the different papers you wrote come together as a whole.
  2. Showing that your work is scientifically robust: The department wants to walk away feeling that they learned a new piece of scientific information, not that they saw an awesome feat of software engineering (at most places). To the degree you are able, framing your talk around a cohesive narrative of research questions, hypotheses, ecological validity, and etc. will help support a sense of scientific grounding, even in the most exploratory empirical work.
  3. Showing that your work is interesting: The job talk is serious, but it should also be interesting and engaging. The best presentations are those that capture the listeners attention and imagination. Explore different ways of opening your talk that will catch people’s attention. For example, start your talk with “Today I’m going to talk about why your grandfather fails to make strong passwords”, instead of reading out the title “Today I’m going to talk to you about Security for All: Modeling Structural Inequities to Design More Secure Systems”. (Credit for this tip goes to Melissa Marshall.) Additionally, make sure that you tell a story - a cohesive narrative. The first 5 minutes of the talk should show why your work is exciting and different and breaks new ground in your area (Credit to Marine Carpuat for this tip). This gives context to those outside your subfield and grabs their interest before you dive into the details. This strategy is also nice to repeat throughout the talk: you want to alternate diving deep into the technical details and allowing the audience to breathe and understand the high-level takeaways from your work.
  4. Showing that you know what to do next: At least 5 minutes of your talk should be reserved for covering your future work. (An even larger portion of your job statement should be devoted to future work.) This is your moment to show that you have plans for the next 5+ years of your research and that these plans go beyond “make a small permutation to my thesis project / existing work.” You should aim to cover two to three threads of research at a high level, but with enough meat to seem tractible (for example, a brief summary of a starter project for each thread).
  5. Showing that you can handle questions: One of the most important parts of the job talk is not the talk at all. Being able to handle questions with poise and in a way that makes the question asker feel like you had anticipated their question but answered graciously will go a long way in improving your job talk. To help with answering questions, it can be a good idea to create backup slides for all potential questions you think you will receive. If you use Powerpoint in presenter mode, you can then type in the slide number of the relevant backup slide when you get that question (make a sticky note that you can leave next to your mousepad with a list of the slide numbers and relevant topics). If you get a question you have not prepared for, don’t stress! There’s no harm in pausing to think, and if the question points out a good suggestion, it’s fine to say - “that’s a great idea, and something I will definitely consider as I expand this work” or etc.  Additionally, you may want to do a bit of research on who is in the department. If you can recognize them when they ask a question this may give you context and a better idea of what you should answer.

Some departments will also have quirky preferences, for example, wanting to see “some math” in the job talk or tending to start talks late but still wanting to end on time. Sometimes your host will also give you tips (e.g., on how to manage hard questions). If you can, check with your advocate/host at a given institution to find out if there is anything unique about job talk preferences in their departments.

It is very important to practice your job talk with many different audiences. This many include your advisor but also people who are outside your immediate area of expertise so they can give you feedback on how accessible your talk is to a general computer science audience.

You can find examples of job talks online. For example, you can find Elissa’s on the Microsoft Research website and you can search for others through the Microsoft Research website or by googling.

How should I schedule on-site interviews? An over-constrained problem.

A theory professor friend of mine (Samir Khuller) said about job interview scheduling: it is an online process and difficult to optimize. The best you can do is try to pre-plan a little bit ahead of time, in an effort toward optimization. You can also try to get as many things out of the way for that semester: for instance, defending your PhD before you start interviewing is a good idea if that is compatible with your committee’s constraints.

You’ll typically have 1-1.5 days of interviews per department you’re interviewing with (multiple departments at the same university means multiple days). On the day of the interview, you’ll be picked up from your hotel for breakfast around 8am, you’ll have breakfast with an interviewer or your host who will, most likely, escort you to the university. From then on, you’ll have back-to-back 30-60 minute meetings with faculty (and, probably, one with graduate students) until 5 or 6 pm, including lunch and 1 hour (or a bit more) for your job talk. You’ll likely have a one hour break, and then go to a 2-3 hour formal dinner with additional faculty members. We tended to get back to our hotel rooms around 10-11pm. Keep this timeline in mind (and especially multiple days of it, back to back, for 1.5 day or multi-department interviews) as you think about scheduling.

Some universities will conduct phone interviews before making on-site interview invitations. These interviews are typically brief (30 minutes) and conducted by multiple faculty at the university. It is a wise idea to look up the professors that are interviewing you to get an idea of their research agenda and seniority in the department. Be prepared to answer questions like: “Do you have ideas of projects you would collaborate with us on?”, “Tell me about something innovative you have done while teaching a course.”, “Why are you interested in this school?”

As on-site interview requests come in, you don’t have to schedule immediately. You can let them know you need a week to firm up your schedule, or if you already have a large number of interviews, you can tell them you’re waiting before scheduling additional interviews. This time will allow you to try to keep geographically close interviews grouped together. If you’re waiting to hear on a school in the same geographic region that you think you’re likely to hear from, consider holding a slot near the slot your scheduling for that anticipated interview.

More broadly, it can be helpful to “hold” a few (2-3) slots for schools you’re especially interested in. We suggest holding slots in the middle of your interview schedule, as if you get offers from your top choices you may be able to stop interviewing, and you’ll be fresher for the middle  interviews than the late interviews, but prepared for your top choices from your earlier interviewing experience.

Also, don’t be afraid to cancel interviews if needed, this is better than showing up and not caring about the interview, which wastes everyone’s time and leaves a bad impression. Related to this last point, aim to do no more than 10 interviews. Even the most extroverted among us will find that more is too much.

When it comes to booking travel and accommodations, we recommend first asking the university if they are able to do the booking for you given constraints that you share with them (e.g., if you have a preferred airline, or you are departing/arriving through an airport that is different). The benefit of having the institution book your travel is that you do not have to claim expenses for reimbursement after the interview is complete, which can save you from learning about each institution’s peculiar way of processing expenses (unless you enjoy paperwork!). If the universities are unable to accommodate your constraints, you can also book your travel directly and ask for reimbursements.

Regardless of how you book your travel, make sure that your itinerary accommodates for flight cancellations and delays, because they will happen and that is a source of stress you do not want to deal with. If you are interviewing in a college town, it can be nice to rent a car to tour the city before or after your interview.

How do I prepare before an on-site interview?

Well in advance of your first interview, you should prepare for common questions you may be asked and have a well prepared set of questions to ask. People will ask you all the time if you have questions. A good starting point for preparing answers and generating questions is this interview question guide, originally started by Soteris Demetriou.

It’s not uncommon to be asked questions that are illegal. For example, questions either directly or indirectly asking whether you are married or have children. We do not have general advice to give on how best to handle these situations, as they are highly personal---but we do want to highlight the need to be prepared as an applicant to handle these questions and reactions. As an example, Elissa was given the advice (from many people) to not wear her wedding ring or mention being married; despite following this advice, she was still asked point-blank during an interview whether she was married. Nicolas was also asked questions that could lead you to reveal whether or not you are married, and it can be difficult to know how to respond to these questions depending on your personal situation. So, be prepared.

You may also want to prepare an adapter kit so that you’re always prepared for your job talk; different departments will have different technology setups and may or may not remember to have a clicker available. For example, you may want to purchase 1-2 VGA/HDMI adapters and 1-2 clickers to keep with you or place in different bags, in case you lose or forget one.

Within one week to 24 hours (sometimes even slightly less) of your interview day, you will be sent a schedule, which will list your one-on-one meetings (and who is attending dinner). It is a good idea to prepare survival guide before each interview. Read about each of the professors you are meeting to find conversation topics and write down the questions you want to ask this person. The conversation topics you select should always include research (you can skim through a few of their papers and have specific questions to ask or areas of overlap to discussion), but can also include questions about hobbies they list on their website, what their favorite activities are in the area, etc.

It is also helpful to write down the name and room number of the person you’re meeting, sometimes your interviewers will forget who you are interviewing with next or not escort you to the next room, so these details can come in handy (and keep you from having to carry around your schedule in addition to your notes). Having a few general backup questions written at the front of your notebook from your broader preparation of questions to ask is useful. If your brain blanks on what to ask, you can easily glance to these notes and use one of your prepared questions. Elissa used a tiny notebook for during the interview day (although she did not take notes during the conversations), which she could slide easily into her coat pocket or purse; Nicolas used a Google Doc that was opened on his phone during the entire interview, and left his phone in airplane mode to not receive any distractions.

Finally, on the day of the interview, remember to grab your adapters, your notes, and carry some water in your backpack!

I survived my interview day. What should I remember after each interview?

It can be a good idea to take notes after each day of interview. When you start receiving offers, this will allow you to remember your first impressions of each university visit. This can help you rank your offers based on different factors that are important to you. Writing these notes after each interview can also help you identify things that you want to pay attention to in your next interviews, as you may realize that factors you had not considered before beginning to interview are going to influence your decision.

I made it! How to negotiate the job package with the department chair?

In advance of receiving any offers, it is a good idea to develop an ideal startup package list (with justification). Many department Chairs will ask you explicitly to share that with them so they can negotiate on your behalf with the Dean’s office. Consult trusted colleagues or even your current department chair to get an idea of reasonable requests, and also ask your advisor or other colleagues for approximate costs for some of the work you led as a Ph.D. student. Having an ideal in mind will help you benchmark the offers you receive and will prepare you in case you receive an offer where they want you to tell them what you want, rather than them making a starting offer. Negotiating your startup budget is important, as you want to have a comfort zone when starting your lab before you rely on grants to fund your research. This is especially true if you are interviewing in countries that have competitive funding calls. This may be less important if you are instead interviewing in countries like Switzerland or Germany where offers include yearly base funding for your entire career. Here is an example template for a startup budget.

In addition to the dollar amount of your startup, there are additional considerations:

Ask questions before making a counter offer / negotiating! Be sure you understand information about money buckets (described above), expiration dates, student costs, etc. Try to do some calculations to understand how your offers equate to each other (e.g., break things into semesters of PhD students, summers of funding, etc.) before you negotiate. Do not feel you need to rush into negotiating (you can generally ask for more time if you are still interviewing for instance), and also don’t feel you need to negotiate with every single university. 

For additional tips and benchmark offer ranges, review this report Elissa generated based on a Twitter-sampled survey she sent out asking for advice about startup and salary negotiation.

How to make a decision?

There are many factors that come into your decision. If you feel like a returning visit would help you, most departments will be happy to arrange that. This can be a good way to meet potential future colleagues you may have missed during your interview, as well as visit the city again, including with your partner or children, this time without the stress of the interview.

At some point, stop listening to advice, and decide based on how you feel.

Good luck!