About the Role
This is a founding-level engineering role at a seed-stage flexible workforce marketplace, sitting directly alongside the CTO to own the entire agent layer end-to-end. You will set the patterns for how every agent in the platform gets built, measured, and improved, working on production systems with real-world consequences at scale. This is a rare opportunity to define the architecture of agentic software in a domain where no one has fully solved it yet.
What You'll Do
Own the full agent layer: context engineering, self-learning systems, evals, the agent harness, and the meta-harness.
Build a compounding knowledge base of undocumented systems so that each new integration gets faster and better than the last.
Design and ship self-healing workflows where agents detect, diagnose, and autonomously recover from production failures.
Build and maintain the meta-harness that lets the team create, evaluate, and evolve new agents in days.
Set engineering patterns that every future agent on the platform follows.
Help hire and grow the agentic engineering team over time.
Work as a peer to the founding team and experiment relentlessly.
What We're Looking For
2 to 9 years of professional software engineering experience, with a hands-on individual contributor focus.
Completed or in-progress engineering, computer science, or quantitative degree from a strong technical university.
Prior experience at an early-stage, venture-backed startup, whether as a founder or an early engineer. A background exclusively in large enterprises or big tech does not fit this role.
Proven experience building agentic software within the past year. This field moves fast and recent hands-on work is essential.
Experience building context or self-learning systems that improve agent performance through accumulated knowledge.
Experience building autonomous failure detection, diagnosis, or self-healing production workflows.
Deep understanding of agent tracing, evals, and eval-driven iteration.
Comfortable building reusable agent harnesses or meta-harnesses for rapidly creating and evolving agents.
Strong implementation focus: you ship working systems in production, not research prototypes.
Proficiency with TypeScript and PostgreSQL; familiarity with LLMs and agent frameworks is expected.
Compensation & Benefits
Salary range: $150,000 to $300,000 USD annually. Visa sponsorship is available where an O-1 visa is a realistic path.
Location
On-site five days a week in San Francisco, CA. Candidates currently based elsewhere in the US are welcome with a concrete, dated relocation plan to San Francisco.