Website:
Job details:
About the Role:
Small cohort. Real work. No fluff.
We're putting together a tight group of AI Safety Fellows to get their hands dirty on problems that don't have clean answers yet, adversarial prompting, evals design, red-teaming, the whole mess of figuring out where AI actually breaks.
No salary. No equity. But there's a real upside.
What you'd actually be doing:
- Running red-teaming challenges and poking at adversarial prompts until something gives
- Getting into papers, arguing about evals, joining research pods when the fit is right
- Competing in hosted challenges, the best performers walk away with cash prizes and fast-tracked job opportunities
- Thinking carefully, out loud, with other people, about where AI goes wrong and why
Who fits:
Hard to reduce to a checklist, honestly. But roughly: you've read alignment or safety papers and came away with takes, not just notes. You've built something, run an experiment, shipped a prompt that surprised you. You don't wait to be told what to explore.
Commitment: Part-time, unpaid, except when you win. Cash prizes and job opportunities go to the top performers in our challenge series.
https://forms.gle/B58HCETivYgAnfDn6
Click on Apply to know more.