- San Francisco, California, United States
- Hybrid
- Full-time · Mid
- $220K–$350K per year
Design and build the backend platform that runs software-writing agents, covering how work is matched to models, how agents receive context, how their behaviour is measured, and how engineers can see what happened. The role combines production engineering with applied research aimed at making these systems better over time.
Compensation
Salary: $220K to $350K - Equity
Responsibilities
- Decide which model handles each task, weighing cost, speed and quality, and measure whether those choices were right
- Shape what information reaches each model by designing lookup, state handling and condensing techniques, keeping only what is pertinent
- Design evaluations that show where to improve next and whether a change made agents better
- Build observability tools that record what each agent did, why, and at what cost
- Use logged runs to improve the system through new tools, cheaper learned behaviour and fixes for recurring failures
Requirements
- Top undergraduate STEM degree
- Nice to have: Program correctness
- Nice to have: Language design
- Nice to have: Model fine-tuning
- Nice to have: Developer infrastructure experience