This is useful. As a technical job seeker, the hardest part with AI-lab postings is often not “is this role interesting?” but “what function is this really in, and who would I actually be working with?”

One bit I’d love to see if you keep maintaining it: a layer for ambiguity / cross-functional weirdness. A lot of postings look like engineering, research, product, policy, and GTM all got blended into one JD. That can be a great role, but it is also where candidates waste a lot of time because the title does not match the actual day-to-day.

Even a simple “function confidence” or “role blend” field could help: e.g. 70% infra eng / 20% research support / 10% product ops.

Nice project — especially because the taxonomy problem is harder than it looks from the outside.

Thanks for the feedback! Re: ambiguity / cross-functional roles: very good point, you're absolutely right about the usefulness of some sort of role blend indicator. I think there's enough data to work on that. Definitely something I'll look into.

Your input as a technical job seeker is very helpful too, gave me a couple of ideas...

Cheers