About the Role
Join an early-stage team building agentic software for real-world business systems. You will own the agent layer end to end, setting the patterns for how agents are built, evaluated, and improved in production.
What You'll Do
Build context and self-learning systems that accumulate knowledge and improve agent performance over time.
Develop workflows that detect production failures, diagnose their causes, and recover autonomously.
Create and evolve the agent harness and meta-harness for building and evaluating agents efficiently.
Establish practices for agent tracing, evaluation, and iterative improvement across the product.
Work closely with the founders, experiment with new approaches, and help hire and grow the team.
What We're Looking For
Two to nine years of professional software engineering experience, including experience building software at an early-stage company or as a startup founder.
Recent experience building agentic software and shipping working systems to production.
Strong experience with context engineering, eval-driven iteration, agent tracing, and production evaluation.
Hands-on experience building systems that learn from accumulated knowledge, autonomous failure detection or recovery workflows, or reusable agent harnesses.
Proficiency with TypeScript, Effect-TS, PostgreSQL, and LLM or agent technologies.
A completed or in-progress degree in engineering, computer science, or a quantitative field from a strong technical university.
Compensation & Benefits
Annual salary range: $150,000 to $300,000 USD. Visa sponsorship is available.
Location
On-site in San Francisco, California, United States, five days per week.