AI Safety Researcher (Fully Remote)

Trajectory Labs, PBC; United States; Remote; USD 83,000 - 152,000; Full time

Job summary

Trajectory Labs, PBC is a startup automating AI safety research. We seek researchers to work on open-ended AI safety problems, producing evals for frontier labs and testing tasks against frontier models.

You will collaborate with PhDs and fellows from Oxford and LSE, joining a small, agile team and reporting to the Head of Conceptual Reasoning. Remote-first with flexible hours (20-40+ per week) and optional Berkeley desk space.

Responsibilities

• Work on open-ended AI safety problems and produce evals for frontier labs.
• Test tasks against frontier models and refine to distinguish strong vs weak answers.
• Collaborate with PhDs and fellows from Oxford and LSE.
• Manage your own schedule with 20-40+ hours per week.

Job description

• Work on open-ended problems in AI safety reasoning, producing evals for frontier labs.
• Test tasks against frontier models and refine until they reliably separate strong from weak answers.

Team:

• Join a small, agile team, reporting to the Head of Conceptual Reasoning.
• Collaborate with PhDs and fellows from Oxford and LSE.

Location:

• Remote-first, async by default, with optional desk space in Berkeley.
• Manage your own schedule with 20-40+ hours per week.

Impact:

• Your work directly influences how frontier AI labs measure and improve safety judgment.
• Contribute to the mission of automating AI safety.

Trajectory Labs

Trajectory Labs is a startup automating AI safety research. We build safety and alignment evals, red-teaming programs, and RL environments for frontier AI labs. As a one-year-old company, we've been featured in major frontier lab safety reports and led high-impact red-teaming efforts.