Reinforcement Learning Engineer

About the role

The team creates deterministic reward functions from real vulnerabilities to train frontier AI models. Work spans fuzzing, program analysis, and scalable ML infrastructure.

At a glance

What you will do

• Design and deploy high-throughput RLVR environments that evaluate LLM actions against execution outcomes.
• Build automated test uses and verification engines to turn vulnerability research into binary reward signals.
• Architect isolated, reproducible execution environments using Docker, BuildKit, or Nix.
• Collaborate with frontier AI labs to define benchmarks and evaluation metrics for cybersecurity tasks.

What they look for

• Understanding of RL training workflows for modern LLM systems.
• Experience with low-level security analysis, fuzzing, or dynamic program analysis.
• Proficiency in building scalable infrastructure for distributed training or high-throughput systems.
• Knowledge of containerization and reproducibility tools like Docker or Nix.

Nice to have

• Familiarity with automated exploitation or patch validation.
• Experience with telemetry and logging for agent reasoning analysis.