Build training, inference, and data systems for SpaceXAI researchers developing frontier coding models.
Scaling existing workAI / ML“We train frontier coding agents https://cursor.com/blog/composer and scale RL on real user data to make them increasingly effective.”
Product areasFrontier coding models · Distributed training infrastructure · Inference and RL infrastructure · Research data systems
GoalsMake coding agents increasingly effective · Speed up research iteration and make progress repeatable · Turn Cursor user data into effective training data
MarketsCursor users
TechReinforcement learning (RL) · Language models · Distributed systems