14

My job search (for tenure-track Assistant professor in Computer Science) last year failed miserably. I submitted applications to about 10 US schools (top 30 -- top-100 quality). In the end, I got only 1 on-campus interview, which I think I did well, but no job offer.

Here is a quick summary of my qualifications:

  • ABD from a highly reputable university outside US (equivalent to, e.g., Toronto, EPFL)
  • 10 papers, all in top conferences (5 first-authors)
  • 4 recommendation letters (one from my advisor, 3 from other reputable researchers)
  • extensive internships in industry research labs
  • extensive TA experience
  • has experience in grant-writing (assisting my advisor)

I am really confident in my research accomplishments and research abilities. I also communicate reasonably well and speak/write English fluently. Since I didn't get a whole lot useful feedback, what do you think are major holes in my profile? I did some reflections and had the following in mind:

  • No Post-doc experience (is it really necessary in CS now?)
  • Never taught a course as an instructor (but I don't see many candidates have this in their resumes)
  • Degrees not from US
  • 10 papers in top venues are still not enough
  • Research/Teaching Statement could be written better (perhaps, but I think mine read OK)

Thanks in advance for your suggestions.

  • 4
    What did your advisor say? – StrongBad Jul 26 '15 at 1:21
  • 6
    Are you holding off finishing your dissertation until you get an offer? That's not usual. Finish your dissertation and then postdoc with your current advisor (or someone else) until you get an offer. – Bill Barth Jul 26 '15 at 1:22
  • 1
    He didn't think postdoc is necessary though — He might be wrong. – JeffE Jul 26 '15 at 1:32
  • 6
    @AleksandrBlekh I would definitely not recommend accepting an instructor position as a viable route to the tenure track. A research faculty position might be okay—I do know of research faculty who later joined the tenure track—but once research is not your job, it's nearly impossible to land a position that primarily values research. – JeffE Jul 26 '15 at 1:52
  • 1
    Whatever the reason that you weren't successful last time, I'd suggest broadening your search this time to include post-doc positions, tenure track positions at lesser ranked universities, and positions outside of the US (You haven't said anything about your nationality, or the place where you've done your PhD, but either could be a negative factor because of simple prejudice or because of real issues with obtaining a visa.) – Brian Borchers Jul 26 '15 at 1:56
18

Let me add two points to user3188445's excellent answer.

Landing a tenure-track faculty position in computer science, even during this current phase of rampant uncontrolled growth, is still very very hard. As user3188445 put it: There's just a lot of randomness in the hiring process. In this respect, hiring junior faculty is no different from every inherently subjective selection process: undergraduate admission, graduate admission, paper acceptance, senior hiring, even tenure.

There is nothing you can do to absolutely guarantee that the dice fall in your favor, but you can change the distribution. As one of my interviewers once said to me when I was looking for faculty positions, the variance is a function of the committee, but the expectation is a function of the applicant.

Apply widely

I submitted applications to about 10 US schools

There's your biggest problem. Only the very strongest PhD students have any chance of landing a tenure-track job with only 10 applications. Each position receives dozens to hundreds of applicants; simple statistics imply that you should aim for a wider range of targets. I'd suggest 40-50 as a more reasonable target.

Hiring protocols vary significantly between American universities. Some departments practice targeted hiring, where every interview candidate must match a specific research area approved in advance by the dean. Other departments practice broad-spectrum hiring; these departments can interview anyone they find sufficiently interesting, regardless of area. Most departments lie somewhere in between these two extremes. Restrictions can change during the hiring season. Moreover, the position advertisement does not reveal the department's hiring constraints, or even necessarily preferences. Some departments publish general ads ("all areas of computer science") even when their searches are targeted; others publish targeted ads even when their searches are broad.

In short, you cannot tell whether your application is in scope for a particular department from the advertisement. For this reason, you should apply to every department that you might be interested in joining, no matter what their ad says. The worst they can do is say no.

Quality > Quantity

My second point is one about perception and language: Your entire application package (including the identities of your references and their contents of their letters) should emphasize quality rather than quantity, impact rather than effort, excellence rather than busy-ness. Even your own internal mental gauges should be calibrated to the strength and potential of your research, not the length of your CV.

At least one of the metrics we use to choose interview candidates is the question "Is this person likely to get tenure?" The more you present yourself as someone who is already ready for tenure, the more your achievements shine on the metrics faculty use to judge each other, the better your chances will be.

ABD from a highly reputable university outside US (equivalent to, e.g., Toronto, EPFL)

The number of non-American CS departments that your prospective American colleagues would rank in the same equivalence class as Toronto and EPFL is very small. Are you sure? Are you sure because Americans have told you so, or only because that's what everyone at your home institution says?

In any case, the reputation of your home institution is not as important as your personal reputation as a researcher. (These are strongly correlated, but they are not the same.)

10 papers, all in top conferences (5 first-authors)

That's certainly above-average productivity, but productivity is not the same as impact. Did your papers win awards? Did they create any buzz? Have they been cited? Did they inspire follow-up work? In short, were your papers actually good? Sturgeon's Law applies even at the best conferences. Do your external letters (see below) describe the novelty, quality, and impact of your work in specific and credible detail? Are you, personally, known for this work, as opposed to your advisor?

4 recommendation letters (one from my advisor, 3 from other reputable researchers)

"Reputable researchers" is too low a bar. You need letters from recognized intellectual leaders in your field, including a significant fraction from outside your home institution. As user3188445 points out, you especially need letters from people who are calibrated to the social norms (as some put it, "gushing") of American academic recommendation letters.

It's also important to have American references so that they can give you direct and brutally honest feedback about your application package from the point of view of your target audience.

extensive internships in industry research labs

Were your internships successful? Did they lead to publishable or patentable results? Did they have significant impact within the company? Did your internship sponsors write you strong recommendation letters?

extensive TA experience

Unless you are aiming for undergraduate-only institutions or a teaching-only position, this is at most of secondary importance. Your research matters first.

experience in grant-writing (assisting my advisor)

Was the grant-writing experience successful? Did you and your advisor actually get the grant(s)?

  • 1
    I agree with pretty much everything you said except "even tenure." A tenure case represents hundreds of hours of labor on the part of the department and external letter writers. People really want to make the right decision. Most tenure randomness comes from the research, not the tenure process. E.g., I've seen someone denied tenure when everyone in the room just really wished we had two more years to see how the research would pan out. But cases are rarely borderline given the facts, which is quite different from wondering if anyone will even read your application to a junior position. – user3188445 Jul 26 '15 at 21:39
  • @user3188445 Yes, tenure decisions require an amazing amount of research and deqliberation and introspection, and the vast majorty of participants in the process really do want to make the right decision. But all the paricipants are human, the set of participants changes from one year to the next, and ultimately, the criteria are subjective. (Even the thresholds for apparently objective criteria—this many journal papers, this high an h-index, that much grant money—are determined subjectively.) With rare exceptions, there is no ground-truth objectively correct decision. – JeffE Jul 28 '15 at 0:54
11

I wouldn't worry about paper count, especially if you have ten papers in top conferences. Do make sure you are realistic about what is truly considered the top conference of your area, of course. But I had only 3 papers in my area's top conference (one first author) when I got a faculty position, and I've placed a student at a top 10 school with less than that. My department gets nearly 1,000 faculty applicants every year. Nobody will have time to read ten of your papers. Much more important is that you have just one thing that you are actually somewhat known for. Then when you apply people will say, "Oh, s/he's the person who did X, I liked that work." Particularly at the junior level, one publication with 200 citations is worth a lot more than 10 papers with 20 citations each. Depending on your area, software artifacts can also be quite helpful if you've gotten others to use them. The number one criterion people are looking for is impact.

Second, don't underestimate the value of the research statement, which in your case you imply was only middling. Somehow people seem to think the statement is not very important. Obviously a truly renowned candidate is going to be interviewed anyway. However, I've served on search committees and seen the statement affect things both ways--strong candidates not get interviewed because of bad statements, and dark horse candidates get invited because of fantastic statements. More than that, writing the research statement is an important turning point in your graduate career. Yes, you've done 10 cool things and gotten 10 top papers out of it... but can you string a sizable fraction of those results into a coherent narrative that shows you have vision? The second most important hiring criterion is potential. You need to convince these schools not just that you've done something cool, but that your accomplishments to date form part of promising research agenda that will continue at their school. I don't think it's possible to put together a good interview talk if you can't first construct a good research statement. On top of that, it makes life much easier for your letter writers if you provide them a copy of a brilliant statement, so indirectly you are probably improving your letters with a good statement.

Third, in many regions of the world, it is customary to write recommendation letters that are less warm than U.S. letters. E.g., I sometimes see letters from Europe that are very dry, like, "The candidate solved X problem. Here are the candidate's strengths. Here are the candidate's weaknesses." Whereas in the US the same person might get a letter saying, "Let me tell you just how important problem X was... And not only did the candidate totally nail the problem, s/he continued to surprise me with a series of follow-on results that seemed intractable just a few years earlier. On top of that, the candidate has been the go-to person for Y and Z in my department, which explains how s/he managed to contribute to fully 10 publications, only half of which I even coauthored. This candidate is as good as Person A (now at U.S. School S), and in several ways stronger than Person B [whom you already interviewed last year]. Our school never interviews its own fresh Ph.D.s, but it sure was tempting in this case--we'll miss the person next year." So when you select a post-doc, make sure to do it either in the U.S., or in a place where people have enough U.S. experience to write the kind of gushing letters we've come to expect here. The people to whom your letter writers compare you must also be meaningful in the U.S. I think most Canadian schools (including Toronto) are similar to the U.S. EPFL (which you mentioned) also has a critical mass of people who understand this, as does MPI. So that might or might not be a problem depending on where you are and who is writing your letters.

Finally, I hate to say it, but there's just a lot of randomness in hiring at the Junior level. There's only one valid way to evaluate faculty candidates, and that is to read their publications (or at least to read the one best publication by each candidate). That takes a lot of time, unfortunately, and typically only happens for a fraction of applicants. Once someone has read and thought about your best paper, your case starts to stand on the merits. Once you actually interview, the randomness goes way down. But what convinces someone to read your paper in the first place? Well, maybe s/he saw a talk you gave at a conference or talked with you there and thought you seemed smart. Maybe something in your research statement or one of the letters caught his/her eye. Maybe you're lucky enough to have worked with someone who has placed students at comparable schools, and so that person can make a convincing case you are worth investigating.

What actionable advice can I give you to derandomize the process? Try to interact with people at schools where you'd want to work and get "on their radar." Serving on program committees or organizing committees might be one way to do this. Talking to people about your or their research at conferences might be one way to do this. But also make sure you aren't annoying. Explaining your latest research result at 400 words per minute to somebody who is tired at the end of a long program committee meeting is not going to win you any points. On the other hand, even if you didn't talk about research, someone could pick up your application and think, "That candidate was really helpful in organizing the poster session, I wonder if his/her research is any good," in which case you are more likely to get your paper read. If you have a friendly relationship with someone at a school where you didn't get an offer, you may also be able to get feedback this way. (Ideally your advisor could work some connections for feedback, but it sounds like that hasn't happened.)

I love my job as a faculty member. On the other hand, I see plenty of people moving from faculty positions at good schools to industry, where they seem just as happy. We are are lucky to be in a field where the most desirable jobs for Ph.D.s consist of far more than just professorships. So I wish you well in your next faculty search, but also take heart in the fact that, whatever happens, you have bright career prospects with a computer science Ph.D.

  • Thank you very much for your insightful advices, professor. It's become more obvious that pedigree does matter a lot, and not being in a top US program is really something I need to overcome in future applications (granted, I was naive to not think too much of it before). It is also a bit surprising to me that 10 papers in top venues per se is not as big a strength as I thought. – justinjamesfly Jul 26 '15 at 18:45
  • 3
    +1 for "The number one criterion people are looking for is impact." Impact is not the same as productivity. Framing your accomplishments as "ten papers in top conferences" is completely missing the point. – JeffE Jul 26 '15 at 19:10
  • Don't get too discouraged, though. You are at a stage where you need to get noticed. Pedigree is one among many ways to increase the probability of someone checking out your application. But even there, what's really important is who writes you letters, rather than where you are earning your Ph.D. Ideally you want letters from people who know you, are calibrated/have clout, and write good letters. It's a bit easier at a top school, where those people are just down the hall. But believe me, even at top-five schools not all faculty members are equally good at this. – user3188445 Jul 26 '15 at 19:11
  • It's a bit easier at a top school, where those people are just down the hall. — Careful. A significant fraction of the letters need to come from people outside your home institution. – JeffE Jul 26 '15 at 19:12
  • 1
    I mean, do you write less positive letters for junior candidates at other schools? — No, but I only agree to write letters for candidates whose work I know well. As you say, the best way for me to know your work well is to work with you. Corollary: Work with people outside your home institution. – JeffE Jul 26 '15 at 20:27
2

Some things committees are looking for which you did not list:

  • Did you win awards? (not directly under your control)
  • Were your publications important? Were they cited? (You cannot control citations, but you can say why your work will be cited.)
  • Can your proposed work attract external funding?
  • Were your applications carefully tailored to each institution? e.g. is your research and teaching proposal relevant to the institution's mission?
  • Experience teaching independently. Designing new classes is even better.

Successful applicants in my field (not CS) have completed at least one postdoc and applied to more than thirty schools.

  • 4
    In my experience in hiring in various STEM fields in the last decade I've never seen a candidate hired at my institution into a tenure track position who did not have some experience post-PhD (either in a post-doc or as a visiting assistant professor or full time researcher at a national lab or...) Although I've never been on a CS hiring committee, I'd be surprised if the market for new PhD's in CS was that different from fields like Electrical Engineering, Math, Physics, Geophysics, etc. – Brian Borchers Jul 26 '15 at 2:01
  • 1
    It is different field-by-field. Post-doc is less common in CS. – Greg Jul 26 '15 at 2:57
  • 1
    I don't think it's worth tailoring the written application to each school. In some cases it can backfire if you are seen as pandering. Once you get the interview, if there's a particularly good synergy someplace it might be worth changing one or two slides to reflect that, but only as the opportunity presents itself, not as a rule. – user3188445 Jul 26 '15 at 9:40
  • @user3188445 around here you will not get an interview if your application is not tailored. You must show that you are not only an excellent applicant, but also that you want the particular job. – Anonymous Physicist Jul 26 '15 at 12:30
  • 1
    I think this varies by both field and department rank. My department just assumes anyone applying really wants the job. We want the best candidates to come and shape our department as much as vice versa. Pandering makes it that much harder to figure out how the person will really be. Now while I'm in CS, I'm in a top 10, not 30-100 department. It's conceivable some lower-ranked departments like the ego boost of a tailored application. But again think of dating: would you date someone who just seems to tell you what you want to hear instead of what s/he thinks? – user3188445 Jul 26 '15 at 17:27
2

Getting interviewed by 10% of the places that you applied to sounds like a success, not a failure. You may just need to apply to more places or you'll need to try a few cycles. If you got an interview than your application was competitive.

  • Sadly, I don't think it works that way. The job market tends to be "winner-take-all." People who get only one interview rarely get the job. I know I said the interview derandomizes things, but there's still a bit of groupthink. People on hiring committees are lazy and too often use other people's judgment as a proxy for their own. "Oh, the candidate is also interviewing at that other school, s/he must be good." (It's frankly a little bit like dating, where you sometimes see person A decide person B is attractive only after B starts dating person C.) – user3188445 Jul 26 '15 at 17:15
  • That said, there's nothing wrong with applying to 20 schools instead of 10. – user3188445 Jul 26 '15 at 17:15
  • 3
    My impression (and CS may be different than math) is that there's a small number of "winner-take-all" candidates who interview everywhere and can take their pick of school (outside perhaps the top 5), but that in the 30-100 tier range mentioned in the post most jobs go to people who had a single-digit number of interviews and no more than 3 offers. – Noah Snyder Jul 26 '15 at 17:28
  • Yes, single digit is fine depending on the digit. Also don't forget that CS has prestigious non-academic positions, too, for which it's much easier to get an interview. So if you interview at three schools plus Google, Microsoft research, and a start-up, that's probably enough to give you an aura of desirability. – user3188445 Jul 26 '15 at 17:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.