Preparing Your Job Talk

Make sure that you schedule a practice job talk of some sort, if only with your local research group. Everyone else will be able to give you advice on it. Make sure that people who are not in your field or office come to your practice talk. During your actual job talks, in general at most one or two other people will be from your field and the rest of the room will be made up of people from other domains within computer science. Besides, the people in your subfield are probably the reason you’re being interviewed; people outside your subfield need to be convinced. The exception to this is industry. In large industrial research labs, for example, you will basically only present to people in your general area.

The general structure or content of a job talk are easily available. We will note that you should definitely include a future work section. If you are lucky enough to be at a department that hires regularly, attend all the talks, even/especially the ones outside your field. Take notes on how much introductory material, related work, detailed material, and future work are presented. Then forget about it for nine months until you have to craft your own.

Although the average is “fifty minutes for the talk and ten minutes for questions”, different places will have different time requirements. At one stop Wes was asked to fit the entire thing into 50 minutes. At another he had 75 minutes. Claire found a similar spread. Prepare your talk in a modular manner so that you can add or remove sections. Bonus points if you can do this on the fly – frequent interruptions are common and can suck up quite a bit of time. Try to avoid giving lengthy answers to questions posed in the middle of your talk (answer the question, but be succinct). At one venue Wes actually had audio-visual difficulties (!) delay (and thus shrink) his talk for ten minutes. Claire recommends you travel with your own laptop and VGA adapter to defend against such hassles. The exception is the FFRDCs (like Lincoln Labs), where defense department rules will require you to email your slides as PDFs ahead of time.

Claire changed her talk modestly between interviews. Moving things around a bit between presentations helped keep the talk fresh. After you’ve done it a few times, the jokes become very unfunny to you. Do what you can to avoid boring yourself.

At some schools, even for a research faculty position, you will be asked to prepare either a short commentary on your teaching style (in 2013, Iowa State requested 10 minutes of the job talk be dedicated to teaching), or an actual teaching demo (Waterloo requested a 20-30 minute example lecture, given separately from the job talk). Claire advises you to not skimp on teaching demo preparation. It can make a big difference in helping you stand out from the crowd, the rest of which probably did skimp.

If possible, do a dry run at a school to which you’re not applying. For example, in September 2012, Claire gave a colloquium talk at Virginia Tech. It wasn’t nearly so intense as a real interview, and it was nice to practice the talk with an unfamiliar audience before doing it for real.

There are different opinions on the “one-third one-third one-third” rule of thumb. A common version suggests that one-third of your talk should be understood by everyone in the room, one-third should be understood by people in your general area (e.g., graphics, programming languages, systems) and one-third should be understood “only by you”.

The trick here is that your job talk must serve many purposes. It must convince them that you are a good lecturer (i.e., the talk must be engaging and speak to your teaching abilities). It will also be their first exposure to your work (as above, many people will be reading you resume at the beginning of your interview – unless you’re at a small department, don’t expect anyone beyond your host to know anything about you or your work unless you say it to them) and must help to convince them that your research has substance. This point is actually somewhat tricky, because if they asked you for an interview they probably already believe your work is good enough (based on you resume, letters, and the local evaluation of the department members closest to your subfield).

It is the personal opinion of both Wes and Claire (thus, no color-coding!) that basically the entire talk should be understandable to everyone in the room. Motivation is key. Remind people of why the entire field of programming languages (or whatnot) is worth considering. A tenured AI professor (as a random example) may well think that “compilers are a solved problem” and it won’t hurt to remind such a person that there is exciting stuff going on here. Claire often used the phrase “This is really exciting because” to help her audience roadmap key contributions. Keeping everyone interested will help to convince them of you teaching potential. In addition, if people cannot understand your talk they have no way to spot potential collaborations. Our job talks were both designed to be easily understood by everyone (heavy on motivation, context, analogies and pictures). We both received many comments from people about just how understandable they were and we also got quite a few offers for collaboration based on them (e.g., graphics and database people were able to see possible fits, not just PL/SE people). Wes specifically asked people after his talk if the lack of “Greek letters” or “complicated-looking material” hurt his case. Typical responses included “it wasn’t a problem because were able to see how intelligently you handled yourself when answering the questions” or “it was fine because we already know how good your work is” or “no, not at all.” However, Claire was advised of the 1/3-1/3-1/3 rule many times throughout her job search. Thus we cannot give you a blanket guarantee that following our approach won’t hurt your employability. We were willing to take that risk to stand up for something we believed in. We encourage you to consider it carefully. Finally, the exception here is industrial research (e.g., Microsoft, IBM): skimp on motivation in such settings – it will just bore them. It probably won’t count against you (because they explicitly mention that they realize that no one prepares two job talks and that the intro is appropriate in academia) but you can possibly earn bonus points by tailoring your talk to those audiences.

Interview Visits

A typical interview is 1.5 days long, depending largely on the size of the department (bigger department=more people to meet=longer interview). You will arrive near dinner time on day X. Someone may or may not meet you at the airport and may or may not take you to dinner with one or two others. You sleep and wake up on day X+1, someone may optionally arrive to take you to breakfast, you head to the venue, and do the interviews. You first meeting (dinner, breakfast, or interview) is typically with your host. After that it’s random. It is completely legitimate to take notes the entire time (we both did). This will be helpful later when you are trying to recall the conversations you had and the people you met. There’s a break for lunch at a nice restaurant with two to four others, then more interviews. This will last until around 5pm, at which point they’ll take you to dinner (unless they did it before, but multiple dinners are not uncommon) and then you sleep. For a long interview (e.g., IBM research, Georgia Tech, UIUC, or large departments in general) things continue the next morning until lunch. You may optionally arrange for the department to set up a visit with a real-estate agent for you (if you’re staying an extra half-day, for example). Despite the fact that this seems incredibly presumptuous (since they haven’t hired you yet), it’s not insulting or awkward. Even if they don’t extend you an offer, getting a feel for that area will give you a broader basis for comparison when you are considering your actual offers. Claire made such arrangements during her second visits.

Almost all of your time will be spent in offices talking to people individually. There are really only three main factors that vary between interview visits. One is the length of each individual interview: anywhere from 30 minutes to 60 minutes is common. The next is how many individual interviews you have scheduled. The last is whether your job talk is early or late. If the talk is later in the day you will have to include a five minute “elevator” summary of your research at each of your individual interviews. If the talk is late and the interview slots are thirty minutes, don’t expect to get anything done. If the talk is early you can use that time to discuss potential collaborations or answer questions they have (and they will have them!) about your talk.

You will probably have a meeting (or lunch) with the dean (or department head or upper-level division manager or somesuch). This meeting tends to be slightly different from all of your other interview chats. At many places (e.g., Purdue) you will hear about exciting developments and expensive multi-disciplinary centers being built. Initially this may well sound irrelevant to you. You should view this sort of thing as a listing of possible collaborations and sources of funding. You may well receive a copy of the department’s “strategic plan”. Other places are of the opinion that you can either spend your time writing strategic plans or you can spend your time doing actual work. Ask the dean where the department is going (invariably the department is growing and going up in the rankings, unless you’re already MIT or McGill, at which point the department is working hard to maintain its number-one-in-the-country ranking) and what has changed in the last few years. Ask the dean to forecast things seven years into the future (you’ll have tenure then). Ask the dean about the hiring plans for the next few years. If all else fails, ask the dean about the retirement plan (it shows long term interest and you probably don’t know anything about them yet as a grad student). Claire met with at least three deans who had reviewed her application and had questions about her research. This was more common when the dean was a computer scientist. Tidbit: as of 2013, the dean at CMU’s School of CS still maintains an active research program (focusing largely on systems). Their conversation was quite similar to a standard interview meeting, and Claire has since learned that the dean in question is absolutely invested in the technical contributions of anyone to whom the school makes an offer. This was non-typical.

Separate from the dean interview is the interview with the department chair (or local hiring manager). This is typically the last interview in the trip. The chair should suggest that they are still in the middle of their interview process but that they will have completed all of their interviews and concluded their deliberations in X days. Get this figure as firmly as you can and write it down. The chair will also want to know what your time frame for making decisions is (e.g., when will you be done with your interviews). Ask why they are hiring. Is it to develop courses? Bring in students? Do more research? Bring in grant money? Grow the department? Ask if they expect to continue to hire in the future, and how much they have hired in the past.

When you are emailing back and forth to set up the interview, arrange for a meeting with grad students if it is not otherwise on offer (most schools arrange such a thing by default). This can be over lunch or in some common meeting room during one of your slots. Be wary of any academic place that won’t let you meet alone with grad students (or undergrads if you’re at a college). You’ll probably get at least three or four students at once; Claire had meetings with up to 10-15. The more people in these meetings, the more difficult they can be, because the candidate has to lead the discussion.

Sometimes the students will have seen your talk and will have (often very good) questions. Claire’s toughest grilling during her interviews by far was at the hands of graduate students. However, much of the time the students will be young (first or second year) and often won’t really have anything to say. Occasionally they will ask tough questions like “what will you bring to this department that we don’t already have?” or “how does your research influence your teaching?” or “what courses will you offer when you first arrive?” or “how will you choose your grad students?”. You should have answers to all of these anyway – don’t be like Wes and get surprised by them in real-time.

Aside from that, you should take time to ask a bunch of questions, such as about their research, the program requirements, their background, and so on. Claire tried to think of issues that mattered to her as a graduate student and ask about them, such as if they felt the faculty was responsive to their concerns or how much their input on this particular meeting mattered in the hiring process. See below for more ideas about what you might ask.

It really doesn’t matter what you wear, so wear what makes you the most comfortable. No one will remember if you wore coordinated separates instead of a full suit. They might remember if you were obviously uncomfortable in your clothes or if you couldn’t walk because of your shoes. If you are comfortable in a suit and have one that fits you and you want to wear it, go ahead. If you are purchasing your first suit since high school to interview, abort. Both Claire and Wes pulled their interview outfits from their existing wardrobes. If you do want to wear a suit of some kind, practice giving talks and sitting around in it before your interviews.

It is basically impossible to communicate in advance how tiring the interview process is, especially since interview season is also prime get-stuck-because-of-snow season. Flying every other day (especially from coast to coast) is exhausting (East Coasters have a real advantage in terms of jet lag). Plus you’ll lose an hour taking the BART to the airport and then getting the rental car and whatnot. With time zone shifts you’ll never really be able to get a good meal in transit (except for the ones they take you to) so you’ll be eating too much airport food. Aim for a salad over McDonald’s, and pack some snacks. Beyond that, though, being “on” for a day and a half can be draining. Aside from these interviews it’s rare to spend nine hours (!) consecutively talking to people about complicated topics, knowing that your future employment is on the line. You must smile and be friendly and in a good humor the whole time. It’s not difficult but it does take more energy than you think. Talking to professors you don’t know for a nine hours is not like talking to your friends for nine hours. Consider taking Dramamine or sleeping pills or somesuch just to ensure that you get some rest on the plane. Claire always travels with ear plugs, an eye mask, melatonin, and a kindle/book to help sleep, as well as a bathing suit and workout clothing to take advantage of hotel gyms when available. Take care of yourself physically. You do yourself no favors if you are so tired and strung out that your are no longer at your mental best.

Interview Questions

The majority of the interview is made up of one-on-one sessions between you and an elder researcher you’ve never met before. Expect the worst-case scenario: you will have to fill the entire 45 minutes yourself by asking questions. Here are some questions (and follow-ups) one or both of us asked (and, where applicable, the canonical cross-institution answer).

• What’s the funding situation like here? Have you had good luck with grants recently? When you were a new faculty member did you get much help with grant-writing?
• Everyone has only great things to say about this place. What’s your least favorite part about working here?
• What can you tell me about the quality of the graduate students here? How did you go about picking your first few grad students? I’ve heard rumors that a bad grad student can be a big drain on a new professor – how do you avoid that?
  • Answer at schools outside the top 10: The average graduate student here isn’t as good as the average graduate student at Berkeley, so you can’t always just take one from the pool and be assured of getting a winner. However, the best grad students here are just as good as the best ones anywhere else – it’s just that the tail of the distribution is longer.
  • Answer at schools from 5-10: The students here are great, but we lose our best candidates to Berkeley/MIT/CMU.
  • Answer at schools in the top 5: The students here are stellar.
• How important is technology transfer or working with the product divisions? What’s the funding model here? What happens if I pick something to work on and no one in development seems to like it? How do I go about making contacts with developers? How much academic freedom is there?

• What’s the tenure process like here? Can I suggest my own letter-writers? Are there mid-point reviews? Is there a “magic formula”? How many people have failed to get tenure in the last 10 to 15 years?

  • Research Answer: There’s no magic formula. Teaching, research and service are all important. Officially you must be stupendous at two of the three and fair at one. However, you can’t really pick research (or teaching) as the one to be bad at. Teaching can only hurt you – we will look at your numerical teaching evaluations and if they are below a cut-off you fail to get tenure. Otherwise we ignore teaching. The amount of grant money you bring in may or may not be a tenure consideration. We care about paper and venue quality, not quantity. One or two journal papers helps. Before going up for tenure you should teach at least one undergrad class.
• Friends of mine who are professors at other universities have had to hide their teaching awards when they went up for tenure. The subtext was that if you had that much time to be spending on teaching you should have spent it on research. What are things like here?
• What’s the teaching load? Does it change after you get tenure? Can you “buy out” of a class? What does that cost? Who decides what you teach?
• What’s it like being a new single/married faculty member here? How did you make friends when you first came to the area? Was it easy to find things to do?
• Are the faculty young or old? Is the university structured and hierarchical? Do decisions come down from on high? How much power do young faculty members have?
• What’s it like to live here? How long is your commute?
• What’s the worst example of office politics you’ve run into here? What do people fight over?
  • Answer: We’re very collegial. We don’t really have office politics. No one gets mad at anyone else. We sometimes squabble over space.
• What’s the general view on programming languages (or whatever) here? Is it a second-class citizen? Is theory emphasized? Practice? Building big systems?

Generally, try to have a set of these questions lined up. For two days, people will repeatedly ask you if you have questions for them. Claire just repeated hers when necessary.

Questions they may ask you:

• Tell me how your interests are aligned with some of the faculty here. Who could you work with?
  • This was by far the most common question. We encourage you to write down in some obvious place (e.g., at the top of your notebook) the name of anyone you think you could collaborate with so that you can answer this question intelligently.
• Are you married? Do you have a two-body problem?
  • See below on the two-body problem.
• What will your CAREER award be about?
• What will your first student’s Ph.D. topic be (in general)? How many students do you plan to have?
• What courses would you be willing to teach? Are you interested in creating new courses?
• How will you divide your time between research, teaching and service?
• What sort of research will you be doing in the future?
  • This is also a very common question. Nominally your job talk will have addressed this, but they may ask again in a more general sense – do you think you’ll be in programming languages (for example) forever, or do you see yourself shifting to systems or graphics? Take this opportunity to distance yourself from your advisor if need be – you want to argue that you can pursue your own work for tenure. In Claire’s experience, the more prestigious the institution, the more emphasis they give to this sort of question, sometimes phrased explicitly as “What do you envision your tenure case looking like?”
• Where do you get most of your funding? NSF? Other government agencies? Corporate funding?

Several of these questions are secretly trying to determine if you have a clue about how the faculty job works, especially regarding the aspects to which graduate students are not always exposed. For example, “have you considered funding sources for your work?” is mostly an attempt to determine if you have the remotest idea how the grant system works.

Wes typically used up almost all of the time asking questions of the other person. Remember, you are there to interview them as well. Moreover, apparently there are candidates who can’t keep a conversation going after the first five minutes, which is the kiss of (awkward) death (Claire heard this from people interviewing her in several departments). You would like to convince the faculty that you are a person they wouldn’t mind working/sharing a hall with. Being able to hold a reasonable conversation for 30 minutes is a good start.

Dual-Career Couples/Divulging Your Partnered Status

Context: Wes was single when he was on the market; Claire had a partner who was planning to move with her and find a job in the software industry; Kevin Leach and Yu Huang had partners seeking academic positions together in the same department. Wes mentioned his relationship status in his cover letters; Claire did not. The advice in this section applies to both regular dual-career couples and those facing the actual academic two-body problem. However, Wes and Claire are less experienced with the latter situation, we cannot provide as much insight on the subject. If you are in that position, we encourage you to talk to others who have done a two-body academic search for more concrete and informed advice.

As mentioned above, you may be asked about your marital status during your interviews. Claire actually never had anyone ask her this outright. Perhaps times changed since 2005, or perhaps interviewers are more sensitive about the issue with a female candidate than they are with a male candidate. Regardless, this question is illegal.

One final note on partners and jobs: emigrating to Canada with a partner appears to be pretty trivial, based on Claire’s (serious) conversations with Waterloo on the subject. The Canadian immigration system is much more straightforward than the US system (as of July 2013), and their HR departments are practiced in providing support for the process.