Product Manager, Agent Harness (Robotics)

What you'll do

• Lead the harness from product vision to focused releases with measurable acceptance criteria.
• Define agent workflows for planning, tool use, context and state management, execution, verification, and recovery; improve reliability using evaluations, feedback, and telemetry.
• Work with technical users and engineers, prototype workflows, integrate tools, and investigate failures through code and execution traces.

What they ask for

• Meaningful ownership of shipped capabilities on a coding agent, such as Codex, Claude Code, Cursor, Devin, Lovable, or a comparable product.
• Deep understanding of agent systems and their design tradeoffs, including planning, tool use, context and state, execution, verification, and recovery.
• Hands-on ability to navigate unfamiliar codebases, prototype workflows, integrate tools, and investigate failures.
• Experience improving agent reliability using evaluations, user feedback, and telemetry, including task completion, human intervention, latency, and compute.

Nice to have

• Familiarity with robotics and physical AI, such as inverse kinematics, motion planning, control, reinforcement learning, imitation learning, VLAs, or WAMs.
• Experience with 3D reconstruction or simulation and robot-learning infrastructure, or familiarity with robot design and manufacturability.

Tools

Codex, Claude Code, Cursor, Devin, Lovable, VLAs, WAMs, CAD

AI focus

Build and improve an agent platform that researches engineering solutions, helps design manufacturable robots, trains robots in simulation, and supports real-world deployment.