RL Research Scientist - Coding Agents & Data-Driven Impact

Cursor; California (MO); On-site; USD 140,000 - 230,000; Full time

Job summary

Cursor is building the future of coding by training frontier coding agents and scaling reinforcement learning on real user data. We seek Research Scientists who can own ambitious, end-to-end RL research in a small, talent-dense team dedicated to truth-seeking and creative exploration.

Expect greater scope and autonomy than typical labs. You will drive hypotheses, design experiments, assemble training/eval pipelines, and push breakthroughs into the next model while maintaining a strong focus on

Qualifications

• Deep background in reinforcement learning (RL) and strong machine learning fundamentals.
• Excellent programming and software engineering skills.
• Ability to tackle ambiguous research tasks with minimal guidance.
• Care about data quality and ability to dive into data when appropriate.
• Truth-seeking, focused on scientific understanding over proving ideas.

Responsibilities

• Own ambiguous, hard research problems end-to-end: form hypotheses, design experiments, build training/eval/data pipelines, and push results forward.
• Improve RL methods for longer horizon tasks and efficient training.
• Train graders to improve performance on coding tasks with non-verifiable reward.
• Improve the quality and difficulty of datapoints used for model training.
• Develop real-time RL approaches for coding agents.

Job description

Cursor is building the future of coding by training frontier coding agents and scaling reinforcement learning on real user data. We seek Research Scientists who can own ambitious, end-to-end RL research in a small, talent-dense team dedicated to truth-seeking and creative exploration.

Expect greater scope and autonomy than typical labs. You will drive hypotheses, design experiments, assemble training/eval pipelines, and push breakthroughs into the next model while maintaining a strong focus on