Researcher, Recursive Self-Improvement Preparedness
Core
Researching and preparing for future AI safety threats, specifically recursive self-improvement, by measuring risks, designing mitigations, and building verification tools.
Role type
Researcher, AI Safety & Risk Mitigation
Builds
Monitors for data poisoning, experiments on model behavior, tools for interpreting model reasoning, prototypes for verifying compliance with future safety agreements.
Domain
Artificial Intelligence Safety, Recursive Self-Improvement, AI Alignment
Deliverable
production ML models | research | infrastructure
Required skills
Strategic reasoning in weak feedback loop domains, rapid prototyping, technical execution, risk measurement, experiment design, stakeholder coordination
Preferred skills
Experience in ML research, AI alignment, AI verification
Technologies
Coding inference systems, model activation analysis, safety monitoring frameworks
Responsibilities
Triangulate trends in automation progress to inform alignment investments, implement defenses against data poisoning, design experiments to assess model misalignment, prototype mechanisms for verifying compliance with future AI safety agreements, prioritize concrete work directions for future security threats
Seniority
Individual Contributor, Researcher