Wednesday, February 1, 2017

Advice on the CS faculty interview

Here is a bunch of advice that people gave me on interviewing for an assistant professorship in CS in 2014 that proved helpful at the time. Please take all of this with a grain of salt (I have never been on a tenure-track faculty hiring committee).
  1. Enjoy it! It's a lot of fun to visit these places and get acquainted with a bunch of interesting people you would probably never talk to. You will keep bumping in to these people throughout your career.

  2. Michael Ernst's advice is good: . Also read my interview with Eugene and Phil Guo:

  3. For the talk: Give MANY practice talks before your first real interview. Take advantage of any offers to sit and watch a practice talk. Giving the same practice talk multiple times to the same audience is not crazy -- you can take feedback, incorporate it, and then give the talk again. Consider videotaping your practice talk and then watching the video. If you can schedule yourself to give a talk somewhere that doesn't matter as much (e.g. a conference, or maybe a lower-ranked school) before the higher-stakes interviews, do it.

  4. Remember that the talk is for a broad CS audience, and the theorists and AI people may have some strange questions based on their own disciplines or experience or what connections they see or don't see. It's good to be prepared for those questions before you get them for the first time at a higher-stakes place, because if you can give an erudite answer off the cuff to some question about a connection to something in somebody else's field (because you've heard it before!) it will make you look like a genius.

  5. Pitching the talk to the audience takes some care. At some schools, you will have an audience of people who really know your area and want to ask tough questions. At some schools, there's really nobody who's an expert, and people just want to get a sense of, "Is this an exciting person with an exciting *future* agenda?" Try to figure out what kind of audience you have in each place [and be ready to switch modes as necessary]. Don't neglect the "future agenda" part of your talk.

  6. Get the $60 Logitech R800 with the built-in timer and 5-minute vibrate warning and the green laser pointer. Practice your talk with the timer and vibrator for timing. Make sure you have a general sense of what the timer is supposed to be at at various points in your talk, so you know if you have to speed up or slow down. Really do aim to finish in 45-50 minutes (including questions in the middle), and be prepared that some places may want less than that.

  7. Have backup presentation options. Make sure you can connect your laptop to the projector at each place. Carry all the dongles necessary. Carry your presentation on a USB thumb drive in various formats (PDF, PowerPoint, whatever) just in case. At Princeton, a colleague had trouble connecting his laptop to the screen and had to delay his talk by 20 minutes and people left before it began and there were no questions and he didn't get an offer. Don't let that happen to you.

  8. Give yourself some rest time in the schedule if you can. If you can avoid it, try not to do two interviews the same week.

  9. It can be overwhelming (especially the places that schedule you for two full days of 30-minute or 45-minute interviews plus three dinners). It's not crazy to take the list they give you and make index cards of the people you'll be meeting with. For each person, do two things: (a) write 3-6 words on the card that help you remember, "who is this / what is their deal?" Trust me, you are liable to forget when it's the 4th graph theorist you're meeting with at School X. (b) Read the CV and be prepared, if there's a lull in the conversation, to bring up, "Hey, here's a problem that youand I could collaborate on." Sometimes this might require going back six years in their publication history to find something you can talk about, but that's okay and people love that stuff. Remember that they are evaluating you as a possible (future) colleague, more than on the basis of your past work.

  10. Soon after you're done with the interview, send everybody you met with a simple two or three-sentence thank-you email, just saying you appreciated meeting with them and enjoyed your visit and maybe that you liked meeting with their students. A lot of people don't do this (unless they are already faculty somewhere) but the pros definitely do it.

  11. At the dinner(s), remember that you are still being evaluated. Limit alcohol and messy foods.

  12. Try to meet with PhD students at every place you interview. (It's fine to ask to put this on your schedule if it's not there in the first version.) You'll enjoy getting a feel for the place through them. Also: at some schools, the Ph.D. students have approximately zero influence (e.g., MIT in my experience) and at some schools, the Ph.D. students have a surprisingly large influence (e.g. Stanford!). You'll never know this in general so just assume the Ph.D. students have lots of influence. From what I can tell, the #1 Ph.D. student evaluation criterion is, "Will this person be a good/kind/helpful advisor?"

  13. Be prepared to discuss how your work relates or is independent from other people at the institution. E.g., if some faculty member is thinking, "What does she add when we already have Professors X/Y/Z?", it could be helpful to situate yourself relative to other people at the institution and show familiarity with their work (and be able to explain it to a nonspecialist).