PhysicsX
AI Engineer: Evals
- London
Builds the evaluation discipline for production AI agents, creating reusable frameworks, datasets and evaluators that combine deterministic checks, expert judgement, model-based grading and human review. Owns cross-product scorecards, public benchmarks and the connection between offline evaluation, regression testing and production telemetry, coaching product and delivery teams to adopt rigorous practice.
Pydantic Evals · Pydantic Logfire · OpenTelemetry