Research Engineer, Safety & Alignment

Sourced from feeny.ai

Cognition, the applied AI lab behind Devin, is hiring a founding Research Engineer for its safety and alignment team. You will build agentic safety evaluations and red-teaming that gate every model and product release, develop alignment methods through post-training, and shape how Devin plans, acts, and escalates in production. This is a hands-on research engineering role focused on making safety work run in the loop of a real autonomous agent.

• Design and run agentic safety evaluations for unsafe actions, prompt injection, data exfiltration, sandbox escape, reward hacking, and misuse
• Red-team Devin and its harness systematically, turning failures into fixes and regression tests
• Develop alignment methods via reward modeling, preference data, and constitutional approaches
• Partner with the agent team on permissions, oversight, and escalation behavior
• Contribute to external safety work through papers, evals, and open methods

• Hands-on safety or alignment research experience (evaluations, red-teaming, alignment, or interpretability)
• Understanding of how agentic systems fail in practice
• Strong engineering fundamentals in Python and PyTorch (or JAX)
• Empirical rigor in experiment design and honest reporting
• Comfort with adversarial thinking
• Prior experience at a frontier AI lab, applied AI company, or developer tools company
• PhD in CS, ML, Physics, Mathematics, or related discipline, or equivalent industry research experience

• Published research in safety, alignment, or interpretability
• Experience building eval infrastructure at scale
• Familiarity with training and inference code