Safety Research Scholar - Human Frontier Labs
Core
Conduct empirical research on the foundations of AI safety, including model failure analysis, safety control testing, and agent behavior study.
Role type
Research Scholar (AI Safety)
Builds
Evaluation harnesses, agent environments, datasets, and monitoring methods for frontier AI systems.
Domain
Artificial Intelligence Safety / Machine Learning
Deliverable
production ML models
Required skills
AI safety research, experimental design, hypothesis formulation, debugging evaluation harnesses, technical writing, reproducible code generation
Preferred skills
PhD in CS/ML/Statistics/Mathematics, experience with LLM agents, alignment auditing, multi-agent safety, recursive self-improvement safety
Technologies
LLMs, agent environments, evaluation harnesses, Python (implied by code generation)
Responsibilities
Design and execute controlled experiments to test AI safety hypotheses, implement and debug evaluation tools, produce reproducible code and technical reports, collaborate with mentors on research direction, submit research papers to top ML conferences
