Research Scientist/Engineer (Agentic Systems)
Core
Build autonomous, large-scale environments to stress-test LLM agents (single and multi-agent) and study their failure modes to improve AI safety.
Role type
Research Scientist/Engineer (Agentic Systems)
Builds
Adversarial environments, multi-agent simulation tooling, and empirical benchmarks for agent safety.
Domain
AI Safety, Large Language Models, Agentic Systems
Deliverable
production ML models | research
Required skills
Agent environment design, multi-agent orchestration, empirical research methodology, red-teaming, experimental design, frontier model fluency
Preferred skills
Published research in automated red-teaming or agentic environments, model failure monitoring, benchmark reproduction
Technologies
LLMs, container orchestration, external APIs
Responsibilities
Build adversarial environments for agents, instrument multi-agent environments to observe emergent breakdowns, run end-to-end experiments against models and APIs, catalogue agent failure modes, turn findings into internal models and public writeups
Seniority
Senior, hands-on IC
