Research Scientist / Research Engineer, Automated Safety
Core
Develop automated red-teaming approaches, safety evaluations, and monitoring systems for autonomous scientific AI models and workflows to mitigate misuse risks.
Role type
Research Scientist / Research Engineer (AI Safety)
Builds
Automated safety evaluation tools, red-teaming frameworks, and risk monitoring systems for scientific AI agents.
Domain
AI Safety, Scientific Superintelligence, Autonomous Scientific Workflows
Deliverable
production ML models
Required skills
Empirical judgment, practical programming, modern AI systems experience, experimental design, failure mode investigation, tool-mediated workflow analysis
Preferred skills
Automated red teaming, adversarial testing, scientific AI, laboratory automation, scalable experimental infrastructure
Technologies
AI models, agent systems, simulated lab environments, evaluation infrastructure
Responsibilities
Develop and run capability evaluations of scientific models and tool-mediated workflows; Create methods to explore diverse failure modes and adapt tests; Build reproducible experiments and evaluation tools; Measure quality and coverage of automated assessments; Develop broader safety automation for threat management and monitoring; Collaborate with engineers to translate findings into safeguards.
Seniority
Senior, hands-on IC
