Put your expertise to work on frontier AI.
Join our network of knowledge professionals who design the hardest ultra-realistic problems today’s flagship models face.
why sherpa labs
We build the RL environments that frontier labs train and test their models on. Our experts design the problems, set the standard for golden answers and find where today’s flagship models fall short.
- paid out to experts in Q3 2026
- $1.1M
- a month earned by our most active experts, alongside a full-time job
- $20K
Paid like the expert you are.
The work takes the judgment of someone who’s done the job for years, and the pay reflects that.
Fully confidential.
We never share your involvement with your employer or name you publicly.
Fits around your job.
Work from anywhere, whenever suits you, and take on as many projects as you have time for.
At the forefront of AI research.
You’ll work directly with our researchers and see how your problems shape the models they train.
What the work looks like.
You’ll take the kind of work you do every day and turn it into problems that today’s best models can’t solve yet. That means the multi-step, messy work where a senior person has to make judgment calls. For each one, you also write the answer and how it should be graded.
What you’ll do
- Turn real work from your field into hard, ultra-realistic problems.
- Write the correct answer and the criteria a senior reviewer would use to grade it.
- Test your problems against frontier models to refine them until they hold up.
What we look for
- 2–10+ years in buy-side investing roles (private equity, hedge funds, etc.).
- Exceptional technical understanding of financial modeling.
- Strong attention to detail and comfort with quick, direct feedback.
Join the network.
Send us the basics and we’ll be in touch.