Founding Engineer, RL Environments (Equity-First)

Role overview Develop reinforcement learning environments that help evaluate shopping agents using realistic shopper questions. The work spans task design, scoring criteria, automated evaluation, and the engineering pipelines needed to generate and run environments reliably at scale.

Responsibilities - Convert real shopper questions into tasks with requirements that can be checked consistently. - Create scoring rubrics that assess whether an answer is accurate, satisfies the shopper’s needs, and is safe. - Build automated graders, including LLM-based judges, and compare their results with human reviewers. - Identify ways agents might exploit scoring rules and improve the evaluation design to address them. - Build pipelines to generate, version, and run evaluation environments at scale.

Requirements - Experience with LLM evaluation, reward design, benchmarks, or AI training data. - Strong Python skills and sound production engineering practices. - Experience comparing automated evaluation scores with human judgment. - Clear written communication.

Nice to have - Full-time engineering experience or experience at an AI training data company.

Benefits and work setup - Full-time, remote role based in the United States. - Equity-first compensation; the source does not specify a salary. - At least four hours of daily overlap with Eastern Time is expected. - The application asks for a GitHub or project link and one sentence about an evaluation or environment you found interesting to build.