Research Engineer / Research Scientist / AI Systems Engineer, RSI
Remote • San Francisco • FullTime
Posted 1mo ago
Remote Work Policy
Fully remote
Employment Type
FullTime
Categories
AI Research Engineer
About the job
The Recursive Self-Improvement (RSI) team at OpenAI is dedicated to building AI systems that accelerate and conduct high-quality research. This involves automating real research workflows, improving research productivity through systems and feedback loops, designing evaluations, and training models to develop missing capabilities. The work spans the full lifecycle of model training, evaluation, and deployment to enable researchers to tackle increasingly ambitious problems. This is a high-ownership role for individuals who thrive in ambiguity, move fluidly between research and implementation, and transform emerging opportunities into rigorous, reliable, and scalable results.
Responsibilities
- Design evaluations for research judgment, hypothesis generation and testing, and long-horizon experiment execution.
- Transform real research workflows and model failures into data and evaluation flywheels.
- Enhance model research capabilities through agent harnesses, synthetic data, RL environments, and model training.
- Build and maintain safe, reliable integrations between models and OpenAI’s research infrastructure.
- Develop research agents, experiment-orchestration systems, and sandboxed runtimes to support real research workflows.
- Create metrics and economic models to assess the impact of RSI on research productivity, model capabilities, and safety.
Requirements
- Research or engineering experience in LLM training, model evaluations, agent systems, synthetic data, research infrastructure, or large-scale distributed systems.
- Strong generalist capable of moving between open-ended research and practical implementation.
- Ability to collaborate effectively across systems, data, model training, evaluations, and other research teams.
- Comfort building and maintaining data pipelines, tooling, and infrastructure for emerging AI capabilities.
- Comfort working on problems without clear definitions or established playbooks.
- Rigorous thinking about scientific quality, research taste, safety, privacy, reliability, performance, and scale.
- Excitement about using increasingly capable AI systems to accelerate meaningful research.