Ask HN: Is there anyone working in ML who regrets not getting a PhD? | Hacker News Reader

Ask HN: Is there anyone working in ML who regrets not getting a PhD?

6 pointscogneuroai8 comments

Want to get a sense of how many people who currently work in the industry regret not getting a PhD to advance their career in ML

View on HN

I've considered a PhD because this is very much a field I want to get into, but I also don't really want to be a traditional software engineer anymore. However, I'm also mid 30s and my undergrad degree isn't CS. To get into a PhD program, I'd probably need to get a CS masters. To get a CS masters I have to take at least a year or two of undergrad classes (expensive and time consuming) just to get to a place where I can apply to a masters program. I'd also have to spend a year or more applying. Between masters program, then applying to PhD program, then finishing PhD program, I would be in my early 40s. I would also likely be taking a pay cut from my product manager career now, and I would lose out on a ton of income over the course of my PhD program (probably close to a million dollars worth).

It just doesn't make sense to spend 7-8 years of school for only 20 meaningful years left of my career when I finish.

Yeah but life is not an optimization problem, and you can't take that extra million with you when you die, so spend it how you want to, not how some algorithm tells you you should. (until ML algorithms take over the world and we're forced to slave away in the code mines for Roko's basilisk, anyway)

If ChatGPT and Stable Diffusion are fun and exciting to you, and you see their potential rather than get mired in their downsides, the least you could do for yourself is to move to product manage projects in that sphere, even without a PhD in the area.

> the least you could do for yourself is to move to product manage projects in that sphere, even without a PhD in the area.

exactly my intention :) I am learning as much as I can in the area, and trying to display that learning publicly, to help me get a role in the industry

PM/Product roles in the industry are extremely uncommon though. Overwhelmingly these places are only hiring engineers and researchers, and I understand why.

I work as a DS lead and have been as DS since 2016. I don't have a master's degree, only a bachelors degree in Econ and a bachelor's degree in Stats. However I code since I'm 14yo, so I'm very capable in the programming side of things.

I have missed carrer oportunities due to the fact that I don't have any further academic achievements besides a bachelors's degree. The small startup might hire you, but the large bank, or more "settled" company, or a large consultancy (big4) won't. That has been my experience so far.

> but the large bank, or more "settled" company, or a large consultancy (big4) won't.

This is definitely not the case for applied roles (data scientist, applied scientist, ML engineer), I know many data scientists in large established companies (local and international banks, consulting, retail) who don't have PhDs.

I graduated in 2020 and was considering doing a PhD in NLP as I had already been doing research at a lab. I ended up joining a series A startup, trying to turn research into products.

I don't regret it much, as I was tired of being a broke student. Also I think the problems you run into in industry are more interesting. However, I do think my skills stagnated a bit and I learned more when doing research.

Definitely don't regret it because I'm more interested in the engineering problems around ML rather than research and I don't want to spend that long working on ML anyway. It gets boring after a few years.

If you want to do research, and want to work somewhere prestigious, and plan to do it for a long time, then doing a PhD makes a lot of sense.

This sums up my experience exactly. I technically work in ML and do build (fairly basic, compared to things like ChatGPT) ML models, but the engineering – specifically as it relates to _data_ engineering – is much more interesting to me.

Additionally, the whole process of building a product around models is, again, much more interesting to me than figuring out the best algorithm to use.

Economics & stats background. Bachelor's only.

Reply on news.ycombinator.com