Software Engineer, RL Data

Highlights

• Build tasks, rewards, and environments that train frontier coding agents.
• Iterate on task sets from traces and evals until models actually improve.
• Scale RL on real user data to make coding agents increasingly effective.

Ideal candidate

Careful, fast-moving engineer who loves turning messy agent behavior into measurable tasks and reusable data systems, with an infra or data background and curiosity for RL.

Requirements

• Write careful, fast code with strong software engineering fundamentals
• Break fuzzy capabilities into concrete, measurable tasks
• Background in infra, data, or distributed systems; RL experience a plus
• Turn messy real-world agent behavior into datasets or tools