Data / ML Engineer, Agent Quality
Make agent quality a measured number, not a vibe. Build the eval suites, quality metrics and regression harnesses that decide whether an agent is safe to ship.
- Build the sales agent's eval harness — labelled scenarios, automated scoring and a pass bar wired into the release gate
- Ship production quality telemetry with a regression alarm when agent quality drops
- Make the eval framework reusable so every future agent inherits measurement for free
Why it matters Deciding what “good” means is judgment; measuring it consistently is the leverage. This is an eval role — we use frontier models as components; our edge is measurement and governance, not model training.
Apply now