JOB-1660
A small, highly technical startup is building tooling that helps enterprises keep their AI agents improving after launch. It links how agents behave in production with the knowledge behind them, so errors can be traced and fixed. The company is hiring an engineer who leans toward the server side but can carry a feature through to the user interface.
The work
- Build the server-side systems behind the platform that scores agent behavior and feeds corrections back
- Shape the schemas that store evaluation datasets, test executions, traces of agent behavior, versions of the knowledge base, issues uncovered and customer-specific quality standards
- Own each enterprise connector from first build to long-term support: authentication, token flows, callbacks, partial and historical data loads, throttling limits, upstream schema drift, retry logic and alerting
- Shepherd each feature from technical design to production release, then watch it, listen to users and refine it
- Join calls with client engineers to trace failures, and turn bespoke asks into capabilities everyone can use
Profile
- Three to six years in full stack engineering with a strong backend tilt, roughly nine-tenths of the work on the server side
- Production ownership of data models, queues or pipelines, or enterprise integrations
- TypeScript for the front end; Python preferred on the server side
- Experience with both a large company and a young startup where you held real ownership
- Computer science degree or equivalent
- Exposure to LLM evaluation or agent QA is a bonus
On-site in the San Francisco Bay Area. Visa transfers considered.
