17
$\begingroup$

My idea of a dream career is a career in industrial research, where one gets to tackle problems which are challenging, and also have a practical usage. To that end, is pursuing a PhD in TCS (I'm interested in topics such as distributed/parallel algorithms, online algorithms) a bad idea ? I've the notion that most, if not all, PhD applicants in TCS are expected to pursue a career in academia - is that a valid precept ?

I've noticed distinguished academics who are regular users of this site - I'm especially looking for guidance from any such person, as I don't want to apply for a doctoral program with the wrong goals in mind. To give a brief description of my profile, I've graduated with MTech in CS (from a fairly reputable institute in my country) about 2 years back, and been working in a MNC ever since on kernel level device drivers.

This question may not be the most ideal question for this site, but I couldn't find a better place to ask it. Requesting the moderators not to close this outright, as I'm really in need of some advice here.

$\endgroup$
  • 15
    $\begingroup$ Pursuing a PhD in TCS is certainly one entry path for your ideal career, although not the only one. And it's certainly not true that all PhD's in TCS are expected to get academic jobs ... lots of them go into industry, and I don't believe there are anywhere near enough academic jobs to accomodate them all. $\endgroup$ – Peter Shor Mar 5 '11 at 16:52
  • 1
    $\begingroup$ Thanks for replying ! Regarding your reference to a PhD not being the only path, I'd agree partially - most, if not all, job descriptions I've come across (entailing research), state they'd prefer someone with a PhD...some even insist on it. Unfortunately for me, I guess that is valid as well - the kind of through grounding to research that a PhD is supposed to provide, can scarcely be matched by a Masters degree. $\endgroup$ – TCSGrad Mar 5 '11 at 17:13
  • 10
    $\begingroup$ I think it depends on what you mean by industrial research. If you want to have the title of "Researcher" at a big tech company like Google/Yahoo/Microsoft, you should probably have a PhD, and a focus in TCS will be in demand (Although perhaps not as much as a focus on machine learning, for example). On the other hand, if you want the title of "Software Engineer" (Which at large tech companies still involve a large degree of independent "research"), then you don't need a PhD, and you should only get one for the joy of the experience, not as a means to an end. $\endgroup$ – Aaron Roth Mar 5 '11 at 20:56
  • $\begingroup$ @Aaron - Interesting distinction that you made, though I'm not very sure that a regular SE would be exposed to much of research (I'm a "System SW Engineer" myself, at a firm which is pretty renowned worldwide, yet I don't even see anyone at my level even remotely associated with anything that can be called "research" - its all development !!). Also, I'm not looking at a doctoral degree as a "necessary evil/means to an end" for such a job, but nevertheless, I would want it to take me closer to what I aspire for... $\endgroup$ – TCSGrad Mar 6 '11 at 4:54
14
$\begingroup$

The comments are all pretty spot on (and maybe they should be answers). To add to them:

In industrial research of the Google/Yahoo!/AT&T/MSR/IBM variety, there are definitely specific area interests, but there's also interest in people with broad interests and strong foundations. A Ph.D in TCS can provide such a broad set of interests and a strong foundation if you choose the right topics.

So for example you could do a Ph.D in computational geometry, focusing on combinatorial problems in incidence geometry, or you could focus on approximate high dimensional algorithms. It's possible that the second topic might allow you to "sell" yourself more effectively as a data analyst, especially if you can point to some awareness of the applications of your work (some papers etc).

So even if you were to do a Ph.D in TCS as a path towards industrial research, the success of this plan really depends on the kinds of topics you choose. In the long term, I think it'll make you very flexible and able to work on a variety of things, but you'll have to seek out that breadth during your education.

$\endgroup$
  • $\begingroup$ Just a thought of mine. I heard talks from some persons worked in IBM research. Although they are really smart (their papers are outstanding, etc), I really doubt their topics benefits the company in any way. I am interested in TCS on my own, and I did applied Ph.D. this year. I really cannot know the connection between their papers and the company's benefits. Sometimes I even think the reason they hire so many bright guys is to prevent their business rivials hire them. $\endgroup$ – Peng Zhang Feb 8 '12 at 18:38
  • 5
    $\begingroup$ @Peng Zhang: When I was at Bell Labs, one of the functions of the PhDs doing undirected research was to serve as consultants to help people in the company who were doing more practical things if they ran up against some really hard problems. If IBM uses this model, researchers can be of enormous value even if their research topics don't directly benefit the company. $\endgroup$ – Peter Shor Feb 8 '12 at 20:09
14
$\begingroup$

One thing to keep in mind is that if you want a job where you actually do research, you pretty much do need a Ph.D. Rarely do big companies these days hire people without a Ph.D. as research scientists. If you decide to do a Ph.D. there are lots of areas where you can possibly focus, and TCS is just one possibility. You can also do a mix of theoretical and applied research or many other options. Whether or not you'll get a job in industry will more depend on their interest in your work (theoretical or not), whether you can sell yourself, the market, etc.

Lots of Ph.D.s in TCS work in industrial research, lots work in academia, and lots leave research all together. There aren't nearly enough academic slots for all TCS Ph.D.s to into academia, so an expectation that you do would be unrealistic and generally isn't there (industrial research jobs are competitive too though). Perhaps your advisor might "expect" you do choose one career path over another, but it's your life and you ultimately get to decide what to do with it.

$\endgroup$

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.