Ask HN: Is getting into a top ~5 CS PhD program possible in mid-30s?

I'm a current graduate student in my 30's in a top five ECE department. I spent 15 years in software but I only recently completed my undergrad.

I'll address only a few things you mentioned.

> Is this even possible, though? Honestly, I know how competitive it is, and while my undergraduate grades are fine and I could nail the GRE again, I doubt I can pass admissions.

Are your grades fine, or are they outstanding? GRE doesn't matter, many CS programs don't even require it. You need to have three strong recommendation letters from people with a background in academic research. Don't even bother getting a recommendation from someone without a PhD. Ideally, the people recommending you have published notable work in your chosen specialization. Once you've got that, then you only need convince one professor to provide funding for you and you will be admitted to the program. Age doesn't matter. If anything, it helps you relate to the professors you'll be interviewing with.

> Also, how do people who go to grad school as adults line up the consulting work?

They don't. Not in a top department. Your research group will look very unfavorably on doing outside, non-research, work. Graduate school is not like a job where you can leave it at the office at 5 pm. It becomes your life. It's unlikely that you'll be able to handle 10 hours a week.

I know quite a few graduate students in their 30's and all of them rely on a spouses income in addition to a stipend. I live with my girlfriend who is also a graduate student. Our lifestyle is not lavish, but money is not a source of stress for us.

Don't be terribly concerned with health care costs. My university provides better health insurance and health services than any company I've worked for.

GRE doesn't matter, many CS programs don't even require it

The GRE isn't irrelevant.

Among top CS programs (top 20), what I've heard is that it doesn't matter as long as you get a decent score. And if you need to do heavy studying to get a decent score, there's little chance your grades and/or research record is good enough to get admitted in the first place.

I did some machine learning research in college. Suffice to say, consulting is quite common in ML. Half the AI/ML professors at my school consulted/advised on the side, and many grad students did as well.

1. Your grades aren't that important. Most important is evidence you can do research (publications or letters of rec.)

2. It is possible to consult in AI/ML, because there is so much industry demand, but you definitely can't count on it. Also, you probably won't be able to consult in your first 2 years, because you'll be busy ramping up and taking classes / passing quals / figuring out your research area. You might have a better shot at relying on internship income (you can make ~30k over the summer as a research intern for companies). But the catch in all of this is that the more consulting/interning you do, the more that will delay graduation (unless you get top tier research papers out of your internships).

3. I would get started with MOOCs and online courses on the fundamentals in whatever area you're interested in, on your own time, and after that start reading papers / implementing things / experimenting. Then some options for you:

a) you might consider approaching professors for volunteer research work (which does happen), or even paid research assistant positions (low pay, obviously).

b) you might consider applying for a top-tier master's first, with a focus on getting and doing research. This will be expensive (like $120k for two years?), however, and worsens the opportunity cost (especially because it is not guaranteed you get into a top PhD after the master's).

The average GPA for UT of accepted people is 3.84:

http://www.utexas.edu/ogs//prospective/stats/pdf/GPA2015.pdf

UT is a pretty good CS school, but I wouldn't say a top school.

I said they aren't that important. Obviously, a low GPA (3.0 or less) will get you rejected unless you have mindblowing letters of rec / publications. But, tons of people have high GPAs, so a high GPA is not going to get you into a top program. Also, an average GPA, but great recommendation letters (and/or publications in top venues) will get you into a top program. Interpolate as you will for GPAs between a 3.0 and 3.5.

Also the kind of person who gets papers published or gets great recs is going to be correlated with a high GPA.

> I'm looking to transition out of startup nonsense and move into basic research.

There are options other than startups and moving to research.

Are you considering those?

It's not entirely clear to me from the post what you're hoping to attain by completing the PhD.

By the way, mid-30s isn't old, 10 years of experience isn't much, and a director-level position in a startup isn't particularly meaningful (if you were a director at a major corporation it might be).

To play devil's advocate: I'm in my mid 30's myself and I'm not a director anywhere, so it does sound impressive to me (ok, maybe not enough for top 5 school ambitions or whatever).

And it does seem old compared to the perception of how life is seen in Silicon Valley / startup culture.

I think you are over glorifying the academic world. From my experience it is pretty similar to the real world. Professors are writing grants and trying to get funding. Its kind of a rat race. Sometimes they get things they like sometimes they don't. It might be easier at a top 5 school to get funding for cool projects.

Also they treat international students like shit. One professor asked me to take a fewer hours so I could spend more hours doing research. However I was not a foreign student and didn't need his funding so I just stopped working with him.

I read recently that CS PhDs are having a very hard time finding jobs. I might be mistaken. That's an assumption you may want to verify -- that getting a PhD will result in the kind of job you want.

Ideally, I'd go to one of the top 4 departments: Stanford, MIT, Berkeley, CMU. Graduates of top universities with decent skills don't have a hard time finding jobs. I'll probably add Columbia and Harvard (middling departments, strong lay prestige) as (irony here) "safety schools". A quant fund isn't going to know the difference between a Harvard CS PhD and one from MIT (or might put Harvard higher).

Anything below the top 20 (and it's top 10 in the rest of STEM and top 1-2 in humanities) has to be purely for the education, though. You're right that a #40 PhD, while the educational program is probably fine, won't fix your career.

Also, the academic job market is terrible and that does not change if you have a PhD from an elite department. That's not what I'm aiming for.

garkimasera's comment seems to be shadowbanned, so I'm re-posting his comment for visibility.

Ideally, I'd go to one of the top 4 departments: Stanford, MIT, Berkeley, CMU. Graduates of top universities with decent skills don't have a hard time finding jobs. I'll probably add Columbia and Harvard (middling departments, strong lay prestige) as (irony here) "safety schools". A quant fund isn't going to know the difference between a Harvard CS PhD and one from MIT (or might put Harvard higher).

Anything below the top 20 (and it's top 10 in the rest of STEM and top 1-2 in humanities) has to be purely for the education, though. You're right that a #40 PhD, while the educational program is probably fine, won't fix your career.

Also, the academic job market is terrible and that does not change if you have a PhD from an elite department. That's not what I'm aiming for.

A quant fund is almost certainly going to know the difference, since it's filled with professional researchers. That being said, Harvard CS is still top 20, and your publication record will probably matter more.

On this point, I wouldn't say that CS PhDs are necessarily having a hard time finding jobs but rather CS PhDs aren't by themselves particular valued in industry so you generally aren't going to get your time and money's worth immediately after graduating (i.e. you'll get paid somewhat less than someone who spent the same amount of time in industry that you spend getting the PhD).

What it can get you is opportunities that people without the PhD don't have available (like corporate research teams that only hire people with PhDs or put them into a separate class called "researchers" vs. other people who are hired strictly to write code).

Whether it helps you longer term in the positions that are available to non-PhD holders entirely depends on you. I've seen plenty of people with PhDs struggle to advance and plenty without PhDs excel while working in the same environment.