Founding Engineer, Environments + Data

About the Role

Build evaluation environments and benchmarks that measure whether AI agents can perform real-world company work and meet or exceed human performance. This founding engineering role combines agent evaluation research with production readiness, helping establish reliable evidence of agent capability.

What You'll Do

• Design synthetic company environments with realistic email, records, and files for agents to use over extended periods.

• Build graders and evaluation systems capable of assessing advanced agents, including cases where human reviewers may lack the relevant expertise.

• Develop benchmarks that correlate with real-world production performance.

• Find and address evaluation failure modes, including reward hacking and benchmark gaming.

• Simulate software tools such as inboxes and CRMs with enough fidelity to support agent interactions at scale.

What We're Looking For

• Hands-on research experience in synthetic data pipelines, benchmarks, evaluations, or model post-training; this is a research-oriented role, not a pure infrastructure engineering position.

• Experience building environments or benchmarks used to train or evaluate agents at a frontier AI lab, and implementing graders or evaluation systems for AI agents.

• At least 1 to 2 years at a fast-growing or AI-focused startup, plus mid-level or senior engineering or research experience.

• Experience simulating software systems for agent tool use, identifying benchmark flaws, and detecting and fixing reward hacking in evaluation systems.

• A degree from a highly selective university or an equivalent credential, such as founding a company or participating in a recognized founder program.

• High-level competition experience in mathematics, programming, debate, or games is a plus.

Compensation & Benefits

Meaningful founding equity is included. Visa sponsorship is available.

Location

On-site five days per week in San Francisco, California, United States.