Research Engineer – Cybersecurity Reinforcement Learning

About the role

Anthropic is seeking a Research Engineer to join its Cybersecurity RL team within the Horizons group. You will help advance the security capabilities of our large language models, focusing on secure coding, vulnerability remediation, and defensive cybersecurity workflows.

Key responsibilities

• Design and implement reinforcement‑learning environments tailored to cybersecurity tasks.
• Run experiments, evaluate model performance, and iterate on research ideas.
• Integrate successful approaches into production training pipelines.
• Collaborate with researchers, engineers, and external cybersecurity specialists.
• Contribute to the safe and beneficial deployment of AI‑driven security tools.

Required profile

• Proven experience in cybersecurity research or applied defensive work.
• Hands‑on experience with machine learning and software engineering.
• Ability to balance exploratory research with production‑grade implementation.
• Strong interest in AI safety and building trustworthy systems.

Required skills

• Machine Learning
• Reinforcement Learning
• Software Engineering
• Security Engineering
• Fuzzing
• Detection and Response
• Capture‑the‑Flag (CTF) competition experience
• Cyber‑range development
• Large Language Model (LLM) training methodologies

What we offer

• Competitive annual compensation ranging from $300,000 to $405,000 USD.
• Opportunity to work on cutting‑edge AI safety research at a fast‑growing company.
• Collaboration with world‑class researchers and engineers in a supportive environment.