Help put AI into production — accountably
KLA builds the runtime layer that lets regulated teams deploy AI agents with policy checkpoints, human approvals, and signed evidence. We are a small, early team backed by NVIDIA Inception and MonacoTech, working on problems that decide whether AI ships in the regulated economy at all.
01What you'd work on
Real surfaces, not a backlog of tickets
These are the parts of the product someone joining now would own a meaningful piece of.
The policy engine
Evaluate identity, risk tier, tool access, and thresholds on the execution path — fast enough to sit in front of a live agent action.
Human decision routing
The Decision Desk: route high-stakes actions to the right named reviewer with the exact context, and bind the decision back into the run.
Cryptographically sealed evidence
Turn every governed decision into offline-verifiable lineage — the record an auditor or examiner can replay years later.
Govern-in-place integrations
Meet teams where they are: instrument existing agent frameworks via SDK and OpenTelemetry without forcing a re-platform.
02Who tends to thrive here
The kind of person this work suits
- You are comfortable in regulated, high-consequence domains — finance, healthcare, the public sector — where being wrong has real cost.
- You care about correctness and evidence: you would rather ship a provable decision than a fast-but-unaccountable one.
- You think in systems and trust boundaries, and you can hold both the engineering and the regulatory framing in your head.
- You are an early-stage builder who is comfortable with ambiguity and wants to own a surface end to end.
How we hire
We are early and hire selectively as we grow, so we do not always have a formal list of open roles posted. If the problems above are the ones you want to spend your time on, we would rather hear from you than have you wait for a job board to catch up.
Tell us what you have built, the surface you are drawn to, and how you think about governing systems where being wrong is expensive.