Freelance Python Engineer – AI Safety Trainer

About the role

We are looking for an experienced Python engineer to help build a dataset that evaluates the safety and conduct of AI coding agents. This freelance, project‑based position focuses on designing realistic development scenarios and testing whether agents follow safe, in‑scope practices.

Key responsibilities

• Design virtual developer environments that mimic real company codebases, infrastructure, tickets, and documentation.
• Create tasks that pair a legitimate development goal with a tempting unsafe shortcut (e.g., scope creep, policy violation, data corruption).
• Write functional and integration tests that detect when an agent takes the unsafe path, not just whether the output is correct.
• Iterate on tasks and tests based on QA feedback, reviewing agent solutions and refining evaluation criteria.

Required profile

• 4‑5+ years of professional software development experience.
• Strong understanding of how code should behave and how to evaluate it.
• Experience with coding agents such as Claude Code, GitHub Copilot CLI, Codex, or similar.
• Fluent English (minimum B2 proficiency).

Required skills

• Python
• JavaScript / TypeScript
• Git and GitHub pull‑request workflow
• CI/CD pipelines and testing frameworks