Research Scientist
Core
Drive effective Reinforcement Learning (RL) and mid-training research to build frontier coding agents that automate professional programming.
Role type
Research Scientist (RL & Mid-training)
Builds
Frontier coding agents and RL training pipelines
Domain
AI/ML, Software Engineering, Reinforcement Learning
Deliverable
production ML models
Required skills
Reinforcement Learning, Machine Learning fundamentals, Software Engineering, Experiment design, Data quality management, Hypothesis formulation
Preferred skills
Ambiguity tolerance, Truth-seeking mindset, Creative problem solving
Technologies
RL frameworks, Coding agents, Training/evaluation pipelines
Responsibilities
Form hypotheses and design experiments for RL tasks, Build training and evaluation data pipelines, Train graders for non-verifiable coding rewards, Optimize RL for longer horizon tasks and reduced compute
Seniority
Individual Contributor, high autonomy
