Research Scientist / Researcher roles @ Big Tech
Looking to switch to Research positions in ML/AI and I hear that the bar has gotten high in places like Meta AI, Google, OpenAI etc. What surprised me is that some of these folks working there seem to be having 5k, 8k citations on their pub record. Is this normal? And is this the expectation when they hire scientists? Requiring 1000s of citations is an insanely high bar IMO especially for early to moderate stage researchers. Anybody have experience with how Recruiting works in these places? What do hiring managers of such research groups typically look for? Citations or top schools or something more? I am an AS with 300 citations in a field that is more theoretical and less applied unlike LLMs but my pubs are all top tier, do I stand a chance? I am also considering applying for postdocs if this doesn't work out but want to get your opinion. #tech #research
Citations hardly matter as long as you’re publishing at the right venues. The bar is high in terms of number of papers published at venues like NeurIPS/ICML/ICLR etc. Having multiple papers at these venues and having a lot of citations makes it a sweet spot but citations is quite secondary. Also, as long as you’re in a relevant field. For example, if you have done plenty of work in theoretical RL for instance, you could definitely apply to teams working on RLHF/DPO and likes of those
It’s more about relevance and quality of your papers and number of publications in top conferences. I recognize the papers in my field and would want to work with the authors of those papers. You should know people in your field through past work experiences and conferences to get recommendations.
I'm looking to make the opposite switch. Research in industry labs is not what it used to. Lots of micromanagement and unrealistic asks (e.g. tell us what the expected improvement is before you have even started to do the work). Any tips from the knowledgeable folks here? My coding skills have atrophied and I don't know how to do the professional development workflow (Jira, Github Issues).
stay there until a well funded startup is looking for a director or staff scientist or staff ml engineer the startup world and ml engineering world is far less pleasant than simply unrealistic asks, its outright pvp throat cutting with almost no practical experience gained, just indians and chinese fighting each other for the scraps that the vcs throw into their cage
Thanks all for your comments really valuable
Looks like there are bad sides to both in the industry: product as well as research side. Unfortunately leadership seems clueless in both!
there are plenty of new grads from ivys at these places, dont let imposter syndrome keep you from putting your best foot forward in your application