Thursday, December 27, 2012

How to get a faculty job, Part 2: The interview

This is the second (actually, third!) part of a several-part series on getting a faculty job in Computer Science. In Part 1, I talked about the application process. In Part 1b, I gave some details about how hiring committees decide whom to bring in for interviews. In this part I'll talk about what it takes to nail the interview itself.

Faculty job interviews are generally one or two (long) days. The main components are the all-important job talk; meeting with countless faculty, deans, and students; and usually some kind of fancy dinner. All of these components are essential to getting a job offer.

The process of interviewing is exhausting. Two full days of talking with people can really wear you out, especially since you need to be "on" all the time. As I'll explain below, any kind of dinner or social outing is not in fact a chance to take a break, since you're being evaluated during those times as well.

Planning travel: Usually, schools will pay for your travel and hotel expenses for the interview, though more often than not they expect you to pay the costs up front and they will reimburse you later. Get a credit card with great rewards since you'll be racking up the points over the course of several faculty interviews. Be prepared to lay out several thousand dollars for each interview trip as reimbursements can take a couple of months to process.

If you are interviewing at several schools, try to avoid doing more than two interviews back to back. Each of these trips takes a lot out of you and it's good to get home to recharge, even if just for a couple of days, in between trips. Also, don't plan on getting any real work done during the interview season. If your thesis committee is expecting a draft, try to get it off your plate before you start interviewing -- that way the pressure is off. By no means should you be trying to meet a paper deadline while interviewing. (Look at it this way: By the time you're interviewing, it's too late for any new publications on your resume to affect the outcome of the job search.)

What to bring and how to dress: You'll be giving a job talk everywhere which almost always means using a laptop to present. Get a lightweight laptop since you'll be lugging it everywhere, and will rarely have a chance to dump it somewhere as you are whisked from meeting to meeting during the interview. Always have your slides -- preferably in a universal format, like PDF -- on a USB stick as a backup in case you can't get your laptop to work with the projector. Also, under no circumstances should you assume that your laptop will have Internet access during the talk -- too many schools have their WiFi locked down and getting guest access can be difficult.

The dress code for job interviews is a topic of much discussion, and I know some people will disagree with me here: But dress formally. For guys, this means a suit and tie, with nice shoes and a nice belt. For women, this generally means a business suit as well, though there is a wider range of options for women who want to dress smart.

Why should you dress formally for an interview? Well, duh, it's a job interview. You want to be seen by your future colleagues as a professor, not just another slacker grad student. You also want to show your potential employer that you are taking the process seriously. At many schools you may have the occasion to meet with a dean or other such muckety-muck who might be the person to sign off on a job offer to you. You want them to see you as a mature professional. I see absolutely no disadvantages to dressing up well for a job interview, and many potential pitfalls for under-dressing.

Yes, you will feel silly at first, since (with rare exception) you will be the only person wearing a suit that you will meet during the interview. People will crack jokes, like "wow! you're really dressed up!" -- my typical response to that was "er, but I always dress this way" which would get a laugh.

It is best to bring two suits and alternate them. You never know when you might spill something on one of your suits, so you need a backup. This also gives you a chance to drop one of the suits off with the hotel to get it dry-cleaned while you're interviewing. Also, always bring your luggage with you on the plane: never check it. You cannot risk your luggage getting lost and being forced to interview in a t-shirt and jeans. I used a nice tri-fold suit bag which was compact enough to hold both suits and fit in the overhead bin on any plane.

The job talk: This is by far the most important part of the interview. If you give a bad talk there is no chance you will recover and end up with an offer, whereas a few botched one-on-one interviews might not sink you. The job talk serves the dual purpose of presenting your research contributions to the department, as well as showcasing your teaching ability. The talk needs to be extremely well-rehearsed, technically solid, clear, entertaining, engaging, and instructive. It is a tall order. If you can't do this well, then you probably don't want to be a professor, since giving talks and lectures is a huge part of the job.

You need to practice your talk, and preferably with an unfamiliar audience -- i.e., not just with people from your research group who already know your work well. Giving a "pre-job-talk" talk at another school is ideal, but be careful: if you blow it there you won't get invited for an interview. Doing a dry run at a school where you don't plan to interview would be a good idea.

It's important to remember that the job talk is not a talk to people in your area. The people in your area (say, systems or AI) already know your work -- which is why you're interviewing there in the first place. The talk needs to appeal broadly to the rest of the department -- to explain why your work is important, what the key contributions are, and to give them intuition for how to solve hard problems in an area other than their own. Don't worry if the job talk feels a little "lighter" than a typical talk you'd give at a conference: You will have plenty of time to get into the hairy details during the one-on-ones.

Margo Seltzer once suggested breaking the job talk into "thirds": The first third lays out the problem space and why it's important; the second third gets into the technical details of your solution (and it's OK to lose some people here, but try not to lose everyone); and the final third lifts back up a level to explain the implications of the work and chart out possible future directions.

As an example, my job talk slides from 2002 are here. I don't want to suggest that it's the best job talk ever, but I think it's pretty good, and got me a few job offers. I always try to have a joke or funny point sometime early in the talk, which helps break the ice with the audience -- for example, around slide 3 of my talk slides I had a funny story about the poor sysadmin of the USGS website not being able to fix his web server for three hours following an earthquake.

Sometimes an interview talk can result in unintended hilarity. When interviewing at MIT, I was asked by Alex Snoeren what impact my system design would have on the "email experience" of a typical user. I responded, "I've never had a mail experience before..."and then suddenly realized the double entendre of what I just said. It took me a few minutes to regain my composure although half the room was cracking up as well.

The one on ones: The bulk of the interview consists of a series of one-on-one meetings with faculty, deans, and sometimes students. These range from half an hour to an hour in length each. You rarely get a break during the day, so if you need to use the bathroom or grab a cup of coffee, just ask (everyone is happy to accomodate). Many of the people on your "loop" will be on the faculty hiring committee, and everyone (regardless of role) will be asked to provide feedback to the committee on whether they think you should be given an offer. So you have to impress everyone. Yes, this is hard to do.

The one on one can take many forms. Usually, you will be asked a bunch of questions about your research, your teaching plans, and future research ideas. You need to spend some time thinking about what you would work on and what kind of research agenda you might pursue as a new faculty member, so you can have a pithy response to these questions. Nobody is going to hold you to it, of course, but you should have at least some half-baked ideas about what would constitute a good research direction when you start the job.

Some interviewers will be trying to assess whether you will be able to get tenure at their institution in a few years. Of course it's way too early to make that judgment during a job interview, but if you can't come up with any kind of coherent research plan or agenda that sounds like it will bear fruit, you're going to be in trouble. When I interviewed, I was doing a lot of thinking about how to apply control theory to the management of complex computer systems, which led in all kinds of interesting directions (few of which I ended up actually working on when I got to Harvard). But at least I had plenty to talk about in terms of possible research directions.

You should also take the opportunity to learn as much as you can about the interviewer. After all, this is not a one-sided process: you should be evaluating the quality of the department and its faculty as well. When prompted, most professors can easily launch into a twenty-minute lecture on their research, so if you find you don't have a lot to talk about with someone, try to get them to do this. You will learn a lot this way and may realize amazing opportunities for collaboration. For example, while interviewing at Harvard, I was really excited by David Parkes' research on multi-agent systems -- and he and I ended up collaborating on a couple of projects once I started there.

The easiest of these meetings are with faculty in your area, since generally you have some common ground. The hardest are with people in completely different research areas. It is a very good idea to cyberstalk your interviewers before the interview, by Googling their names and learning as much as you can about their research beforehand. You might discover that there is some mutual interest or acquaintance this way, which will give you something to talk about. If you don't know who will be on your loop, ask your host and they can usually send you the schedule in advance. It's impressive when a candidate comes in having done their homework, knowing a bit about the interviewer's research and background. This is not creepy (although if you get into how cute their kids' pictures are on Facebook, you've probably crossed a line).

You will invariably meet with someone who was unable to make your job talk, so be prepared to give a 5-to-10 minute rundown on your research, a "mini job talk", if you will. You need to have a punchy, clear way to answer the question, "So, what do you work on?" My opening line was something like, "I work on making web servers really fast, and able to stand up to massive overloads." This was enough to get a conversation going on the topic and was a problem statement that pretty much everyone could relate to. If instead I had launched into, "I work on a hybrid event-driven-threaded server architecture combining rate-limited queues and feedback-controlled thread pools", I would have immediately put about half of my would-be interviewers to sleep.

There are, of course, some tactical questions you should try to get answered while you interview. The standard questions that candidates ask revolve around the teaching load, size and growth trajectory of the faculty, what new areas or initiatives the department might be starting up, what class sizes are like, whether there is a big Master's program, what the department's relationship is with the rest of the school, and of course what the tenure process is like. The interview is not the time to ask questions about compensation or benefits: Save that for once you have an offer (which will be the subject of the next part in this series).

You also want to learn as much as you can about living and working in whatever city the school is in. If you're thinking about buying a house or having kids, you need to understand about the real estate market, schools, good neighborhoods, commute, and so forth. If you care about eating and drinking out, you need to learn about the nightlife. If you ask no questions about the city or area, your interviewers will pick up on this and assume you're not that serious about moving there. You can also save these questions for a second visit after you have a job offer in hand, but it's probably a good idea to start learning about your potential new home.

The dinner: Most departments will take faculty candidates out to a fancy dinner somewhere. This might sound like a real perk, but believe me, after 8+ hours of interviewing, it's usually the last thing you really want to do. A nice glass of wine (or three) might sound like the perfect antidote, but it's probably a bad idea to drink -- you are still being evaluated over dinner, and if you're like me, you can get really uninhibited with the combination of interview exhaustion and alcohol. Of course, for the faculty dining with you, they are planning on expensing the dinner and wine, so by all means encourage them to order whatever they like (and maybe indulge yourself half a glass to help take the edge off).

The best interview dinners I had were with folks that I was friendly with and worked in my area. Dan Wallach at Rice recognized that I was probably getting sick of fancy restaurants and took me out to eat crawdads with my hands (and a big old plastic bib to protect my suit). The worst interview dinners I had were when several senior faculty used the time to gossip amongst themselves and completely ignored me. On that topic, don't gossip about other schools while you are interviewing. It's bad form, and an easy trap to fall into -- and keep in mind that everybody talks to everybody, so what you say at UCSB will get back to those folks at Duke, somehow (not that I would ever do such a thing).

After the interview: When you get home, or back to your hotel, be sure to send a nice thank-you note to your host, expressing your interest and enthusiasm for the school and department (assuming, of course, that you are enthusiastic and interested). Don't assume the school knows you really had a good time and would love to work there. Hiring committees are always trying to read subtle signals from the candidates about how seriously they would entertain an offer from their department, so if you're not explicit, the hiring committee might mistakenly assume you wouldn't be that keen on a position there. If you're not that interested, well, don't go out of your way to say that you are, but you probably don't want to let the school know right away. Having several offers -- even from schools you're not serious about -- can be a good bargaining chip when it comes time to negotiate the offer with the school you do want to join.

Finally, I strongly recommend taking detailed notes on your interviews, when you get back to the hotel each day. I found my notes to be invaluable when considering the several job offers I had, since my memories of a place started to fade after ten or so interviews. Writing out my observations and gut feelings about a school also helped crystallize the many tradeoffs in my mind.

After this it's mostly a waiting game to see if you'll get an offer. This can take a matter of weeks, depending on when during the interview cycle your visit happens to fall, so be patient! If you do end up with a time-limited offer from another school, it's perfectly acceptable to contact other schools you have not heard back from yet to let them know you are still very interested but are operating under time pressure. Stay tuned for the next part of this series where I'll talk about the process of negotiating offers.


  1. An excellent summary of all the do's and don'ts. I think you literally covered all the things I remember as well. So instead of me telling my students this stuff year after year, I'm just going to send them to this page.

    On a side note, what happened with you and UCSB/Duke?

    I'll add a few more tidbits, FWIW:
    - These days, I see many more 1.5 day and even some 1 day interview trips, probably due to increasing # of candidates per spot per school. Some schools are going on hiring binges, e.g. hiring 4 or 5 spots in the same year, so they're reducing per candidate time to reduce interview fatigue for the faculty.
    - I remember folks telling me before I went on the market, AVOID messy meal choices on the interview trip. Spaghetti or similar foods tend to make a mess, and that combined with a dressy outfit = what you don't want on an interview trip.
    - I've always heard of the "hourglass" model of the interview talk, similar to Margo's 3x 1/3, except the parts are less well defined, and the middle part has love handles
    - Hopefully you have the problem of getting a bunch of interviews, in which case I'd highly recommend trying to schedule them intelligently to a) maximize your chances of getting (near-)simultaneous offers and b) not wearing yourself out by putting too many schools together in a row (i.e. sprinkling in some labs will change things up and keep you fresher).

  2. The UCSB and Duke thing was a joke - as far as I know I never gossiped about (or offended) people at either place :-)

  3. Oh, and I strongly encourage asking key questions to multiple faculty to see what kind of different answers they give you. For example: asking younger and senior faculty "what does it take to get tenure at xxxxx" can get you some interesting answers.

  4. I would strongly recommend against wearing a suit to a computer science faculty interview.

    Since it is rare for young researchers in our fields to wear suits to school, it reeks of phoniness. (I might make an exception if someone wears a suit regularly to work and is very comfortable in a suit.)

  5. Doug, of all of the faculty candidates I have met with and interviewed over the years, I reckon 95% of them wore suits. Are you seriously suggesting that this is not the norm?

  6. My experience interviewing faculty candidates has been completely different.

    I would say about 15% of CS faculty candidates wore suits. Especially among junior faculty candidates (with a few exceptions), they looked uncomfortable, and I think it adversely affected their performance. When the thing the interviewer remembers was "wow, that candidate looked really uncomfortable in a suit" rather than "wow, that candidate was really smart," I think it is safe to say there has been a failure!

    Maybe Harvard is different, but I don't remember seeing such a high fraction of candidates in suits when I was a grad student there.

  7. Thanks for the post Matt.

    I think Doug is suggesting that faculty candidates normally look phony...

    While I didn't interview as many places as Matt, I didn't personally find the interview process exhausting. It took a lot of energy, certainly, but to a large extent I think getting to chat with a bunch of smart people about research (theirs and yours) has to be something that you enjoy if you want to be a professor.

    The advice to google the people you have one-on-one meetings with is useful. It's quite uncomfortable if you sit down with someone and have nothing to get the conversation started. So I'd have at least one question prepared for each person (something, at least very high-level, about their research). This does take a lot of time to prepare, however, so waiting until late the night before to start googling is probably suboptimal.

  8. I did a talk, much like your blog series, right after my own job search. I had a slide about being funny and got a lot of push back from the Princeton faculty. Suffice to say that you shouldn't be funny if it's not natural for you. If you are planning to be funny, just keep it clean and keep it focused on your work. I had a recurring example in my talk that riffed on then-President Clinton and why you should or shouldn't trust what you read in the press. Worked for me, but not for everyone.

    Clothing: I wore a suit and tie. On two day interviews, I'd wear khakis and polo shirt on the second day. For an academic interview today, I'd recommend either, but nothing less formal.

  9. Doug - that is amazing. I guess Berkeley and Harvard are pretty different in that regard. Still, I don't think a suit is phony. People should learn how to wear a suit and feel comfortable in it. It's not that hard :-)

  10. Dan - I agree that you shouldn't try to be funny if you can't pull it off. But the talk should try to have some degree of entertainment. A very dry technical talk is a deal-killer for me.

    The best job talk stunt I ever saw was when Nickolai Zeldovich rebooted his laptop (which was running the experimental HiStar operating system he was describing in his talk) ... in the middle of the talk. It rebooted and immediately resumed back to his talk slides. It was a great stunt.

  11. Matt, thanks for detailed and useful posts!

    You say that "Journals don't count either. (This varies by field. In Computer Science, journals don't matter very much" ... even taking the context into account, I would suggest this is misleading. A recently completed PhD in computing typically won't have any journal publications, but that doesn't mean that journal papers don't count. More broadly, journal papers certainly do count in computing, and in recent years there's been a definite trend of increasing emphasis on journal papers.

    I also note that your posts are (understandably) US-centric. It might be worth noting that some practices do vary around the world. For instance, the US style PhD (including coursework) contrasts with the British style PhD (which doesn't have coursework), similarly the way TAs are used to deliver teaching, and the US system of tenure are not universal.



  12. I think the issue of dress is overwrought… perhaps because it’s one of the few things that is easy to control. As a general rule you can’t wear jeans and a t-shirt, but a suit is absolutely not necessary for men. I say wear something respectful, but that you’re comfortable wearing. For the record, I didn’t wear a suit (nor a jacket, nor a tie for that matter) and it didn’t seem to cause any trouble for me.

  13. One thing that hasn’t seem to come up is the issue of affect and personality. I think it’s worth realizing that in all of the one-on-one meetings, dinners, etc… there is a kind of friendship test going on. This is not in any official criteria that any school publishes, and we could debate if it should be involved in decision making or not, but as a practical matter everyone is implicitly judging whether or not you’re someone they would like to have around all the time, would go out to lunch with, might collaborate with, etc. If you come across as distant/withdrawn or arrogant/self-centered, this can strongly undermine your chances. True story: quite a few years back there was a top candidate, from a top program in their field, who had “walks on water” letters, had done very strong work and their talk was technically impressive and easy to follow. They got interviews everywhere. But the person came across like a total a*hole and received zero offers anywhere. I have plenty more stories like that (although perhaps not as extreme in outcome). This is not unique to faculty jobs BTW… as a rule everyone wants to hire people who are positive, excited about what they do, interested in what other people do, engaging and open-minded.

  14. Huge +1 to Stefan's comment about personality match. At Harvard, owing to the small size of the faculty, and the strong desire for everyone to collaborate, we would often pass on "walk on water" candidates who were seen as unlikely to get along with or collaborate with other faculty. We'd much rather have someone who could mesh well with the department than someone who would just sit in their office and crank out papers. Not all schools seem to have this criterion, however - it depends a lot on the personality of the department.

  15. Michael - It's true that all of this is really based on the US model. I don't know how things work in Europe, which I take it uses a very different approach to hiring than the US. Also, this is really only about hiring at research institutions, not teaching colleges.

  16. Very nice post, Matt! I wish something similar had been available when I was interviewing 20 years ago. Here are some thoughts from having been on both sides of the interviewing process.

    In general, try to put yourselves in the shoes of the people doing the evaluation. Consider what you would be looking for if you were on the hiring committee. Even fresh PhDs have seen many hiring talks and (probably) met with candidates. Reflect on what candidates most impressed you and why, then be that person. Get people excited about your work, which starts by being excited about your work. Convey confidence, but not arrogance. Have a compelling story for what you want to work on if you are hired and why it matters. As Matt says, nobody will hold you to it.

    Stefan's point about personality is very true. Interviewers want to hire somebody who they may work with for the rest of their career. Some interviewers do not care about personality and will evaluate you only on technical talent---these people tend to be the ones who themselves score poorly on personality.

    Margo's rule of thirds is good. You need to convince the room that you are a capable researcher and teacher. People from outside your area will tend to defer to the judgement of the area experts, but help those experts make the case for you, rather than work to overcome objections. One mistake that some candidates make is to treat their interview talk like a one-hour history of their research, rather than a coherent message of research accomplishment and a vision for its future. Don't just staple several conference talks together with minimal integration. Convey a coherent and compelling vision.

    The top piece of advice I wish I could give to the 27-yo version of myself would be to schedule no more than two interviews per trip. I literally received offers from every place I visited first or second in my trips, and none from places later. Interviewing is exhausting, physically and mentally, and by the third place on each trip, it is easy to slip into survival mode (How long until I can end this?) rather than presenting yourself in the best light. Spend a weekend at home between trips to rest and recharge and put all other commitments aside.

    Ask to meet with graduate students. You might need to defer this until the second trip, but you will probably learn a lot more about the department from the (frank speaking) graduate students than from most faculty.

    My thoughts on dress are between Stefan and Doug's. Most candidates do not come across well in a suit and tie---as Doug says, they look phony. But some people come across very well in a suit. It's a matter of what projects a stronger air of competence and confidence. If you are unused to wearing a suit a tie (most candidates), you will look phony. If, however, you have a well-fitting suite and wear it like you are completely comfortable in it, go for it. Funny story: Due to a mishap with the dry cleaner, for one interview I was forced to wear tan slacks and a polo shirt, rather than a suit. In the early 90's that was more unusual than it is today. At the end of the day, my host commented that several faculty had remarked to him how confident I must be to dress so informally. I had to swear him to secrecy when I admitted the truth.

    Re: Food and drink. Again, Matt is spot on. I definitely appreciated the casual dinners after weeks of fancy Italian and French cuisine. The best was a home-cooked meal (ok, it was at a penthouse apartment in Manhattan, but...). Just because your hosts are guzzling wine or ordering heavy desserts does not mean you must. And do not assume that just because the conversation through the salad, appetizer, and entree courses has been light and casual that one of your hosts won't decide to crank up the heat with some deep questions come dessert (the MIT trap, as I came to call it).

    Again, great post, Matt. Potential candidates would do themselves a great service to read and re-read it.

  17. Thank you for sharing this, Matt.


Startup Life: Three Months In

I've posted a story to Medium on what it's been like to work at a startup, after years at Google. Check it out here.