Reinforcement Learning Engineer
Core
Design and scale automated verification pipelines that transform real-world software vulnerabilities into deterministic reward functions for next-generation frontier AI models.
Role type
Senior IC reinforcement learning engineer (RLVR)
Builds
High-throughput RLVR environments and automated test harnesses for AI agents to discover and remediate software vulnerabilities
Domain
Cybersecurity + Reinforcement Learning
Deliverable
production ML models
Required skills
Reinforcement Learning from Verifiable Rewards (RLVR), Python, C, Rust, software vulnerability analysis, binary exploitation, fuzzing, dynamic program analysis, DevOps pipelines, Docker, BuildKit, Nix, Linux systems, low-level debugging, benchmark environment design
Preferred skills
Custom reward function design, ground-truth verifiers, automated grading engines, low-level program analysis tools, compiler instrumentation, competitive cybersecurity benchmarks (CTFs), open-source AI evaluation frameworks
Technologies
Python, C, Rust, Docker, BuildKit, Nix, GitHub Actions, ASan, MSan
Responsibilities
Design RLVR environments evaluating LLM action sequences; Develop automated test harnesses converting vulnerabilities to binary rewards; Integrate Mayhem platform into RL pipelines; Architect safe, isolated execution environments; Collaborate with frontier AI labs on benchmark formats; Implement telemetry and ground-truth verification algorithms; Build low-level instrumentation tools; Optimize infrastructure for distributed RL training; Benchmark AI model performance on security challenges
Seniority
Senior, hands-on IC