Offensive Cybersecurity Engineers: Build RL Environments for AI Evaluation

What You'll Do6

• Turn offensive security scenarios into working reinforcement learning environments that AI models must navigate.
• Harden existing benchmarks so they hold up against advanced model interactions.
• Write evaluation criteria that separate real model capability from guesswork.
• Iterate on security scenarios with the research team across the life of the project.
• Refresh environments as new vulnerabilities and attack techniques appear.
• Commit to ongoing hourly collaboration over a long-term engagement.

Requirements6

• Professional experience in offensive cybersecurity or penetration testing.
• Strong background building or hardening technical environments.
• Comfort with long-term, remote, hourly project work.
• Real interest in AI safety and how models get evaluated.
• Ability to design and implement complex cybersecurity benchmarks.
• Willingness to collaborate and iterate rather than hand off work once.

Preferred Qualifications2

• Familiarity with reinforcement learning
• Experience with AI evaluation frameworks

Who Should Apply

Offensive security engineers, penetration testers, and security researchers who already build or harden technical environments. Candidates who have shipped security benchmarks or test setups will settle in fast. Some exposure to reinforcement learning or AI evaluation helps, though the team does not require it.

Salary Insight