Quick Overview
Job Description
About the Role
Join an early-stage AI team building autonomous agents that can carry out complex work over extended periods. You will own core agent systems and evaluations, helping make these systems reliable enough to operate with minimal human oversight.
What You'll Do
Build agents that sustain execution toward a goal over weeks without continuous human intervention.
Design memory systems that help agents retain and build on knowledge over time.
Develop evaluation frameworks to assess agent performance against human-level baselines.
Improve agent reliability and reduce the need for human review of each run.
What We're Looking For
Hands-on experience building dependable agents and a practical understanding of their current failure modes.
Experience developing memory systems, long-running execution capabilities, or multi-step reasoning systems that maintain goals over time.
A track record building evaluation frameworks that inform real decisions and compare agent performance with human baselines.
Published research that others build on, or software shipped at meaningful scale.
Background in competitive math, programming, debate, or games, founding experience, or participation in a leading accelerator.
A collaborative, kind approach, including when work gets difficult.
Compensation & Benefits
Base salary range: $150,000 to $300,000 annually.
Location
On-site, five days per week in San Francisco, California, United States.