New
Research
Research Engineer, Evaluation
San Francisco
Full time
Measure what good support looks like and turn it into evaluations we trust.
About the role
Measure what good support looks like and turn it into evaluations we trust. You will join a small team that ships every week and talks to customers directly.
What you’ll do
- Build evaluation sets from real, anonymized conversations.
- Design metrics that track resolution and quality, not just accuracy.
- Run experiments that decide which agent versions ship.
What you’ll bring
- Experience training or evaluating language models.
- Strong Python and statistics.
- Comfort working from messy, real-world data.