AI Model Behavior Lead | Up to $300K Base, $500K OTE — San Francisco, CA

"Extraordinary claims require extraordinary evidence." Carl Sagan

Everyone has a take on AI models. Very few people can back theirs up with evidence. If you can, keep reading.

Our client is a unicorn AI data company whose datasets and evaluations sit behind the most capable models in the world. Almost every frontier lab buys from them. They're adding two people to a brand-new function in San Francisco, and the brief is simple: be the researcher that lab researchers want to talk to.

THE JOB

You sit with the commercial team, but you think like a researcher. You form sharp views on how models should behave (what makes a personal assistant helpful, an enterprise agent trustworthy, a chatbot worth using) and you prove them. Then you turn that thinking into proposals and analyses that lab researchers read, respect and act on.

You begin in the background, building and writing with the team. In time, you take the room yourself.

HOW CANDIDATES ARE SCORED

The hiring team uses one scorecard for everyone. Here is exactly what's on it.

Must-haves

• Frontier pedigree. You've worked at a frontier or neo lab, or at an AI evals or data company such as OpenAI, Anthropic, Scale AI, Snorkel AI, METR or Redwood Research.

• Hands-on model work. Model behavior, evals, post-training, answer quality or interpretability, done by you personally.

• Clear writing and confident presenting. You can put a proposal in front of a lab researcher, then walk a room through it live.

• On-site in San Francisco. Five days a week in the Financial District, with relocation help if you need it.

Expected

• Around a year or more in model behavior, interpretability or evals. Less is fine if it was the real thing inside a frontier lab.

• Opinions with substance. You can say what makes a model good or bad for a specific user or use case, and why.

A bonus

• You've contributed to benchmarks, datasets or evaluation projects, or worked with human annotator and expert pools.

• A technical degree (computer science, machine learning, statistics, data analytics, economics) or equivalent hands-on experience.

Less likely to fit

• Product-facing roles with no hands-on model work.

• AI experience only at a large cloud provider, with no lab time alongside it.

WHAT YOU GET

• Up to $300K base, up to $500K on-target earnings, plus equity.

• Influence across nearly every frontier lab at once, rather than one model family.

• A founding seat in a new function, with room to grow it.

• One of the deepest ready-made data catalogs in the industry behind your work.

• Relocation support. Existing US work authorization is strongly preferred.

zdv EFj7 BDzKgQy: YOUR TRUE PARTNERS IN AI AND TECH ROLES

We're a specialist AI and technology recruitment firm with a deliberately small desk. You'll deal with someone who knows this client, understands the work and will tell you plainly how you measure up against the scorecard. We prepare you well and back you from first chat to signed offer.

APPLY

Apply here on LinkedIn. Tell us where you worked, what you did yourself on model behavior, and one claim about model behavior you can back with evidence.