Head of AI Education at Cursor, teaching developers how to build with AI coding tools. Previously led Developer Experience at Vercel, growing Next.js to 1.3M+ active developers and helping scale Vercel to $200M+ ARR. Educator focused on making technology easy to understand through courses, videos, and essays.
Champion of Speed & Specialized Models
Lee Robinson represents a new generation of AI product leaders who prioritize developer experience and specialized model design over one-size-fits-all solutions. His work at Cursor demonstrates that faster, domain-specific models can match frontier intelligence while dramatically improving user satisfaction.
Current Work
At Cursor, Lee leads educational initiatives for developers learning to code with AI tools. His focus areas include:
- AI Education - Teaching fundamentals (tokens, context, agents) through six-part video series and workshops
- Specialized Model Development - Building Cursor Composer, a fast frontier coding model trained with RL
- Developer Experience Design - Optimizing for speed and workflow integration over raw capability
- Hybrid Architecture - Using smart models for planning, fast models for execution (“rip through the code”)
Lee writes about developer tools, AI workflows, and web development on his personal site and Substack, with key essays on Developer Experience, Developer Marketing, and Developer Relations.
Background
Previously at Vercel, Lee served as VP of Developer Experience, leading Developer Relations and Documentation teams. He built one of the largest technical communities around Next.js and React, establishing himself as a leading voice in modern web development through extensive teaching, conference speaking, and open-source contributions.
Holds expertise spanning web development, JavaScript/TypeScript, React, Next.js, serverless/edge computing, and AI-powered development tools. Known online as “leerob” across platforms, with 15 years of coding experience and a second half spent teaching.
Philosophy on AI Coding Tools
Lee’s approach challenges assumptions about AI model deployment:
Speed is a feature - Developers prefer fast, responsive tools over marginally better but slower alternatives. Speed measured in seconds, not minutes, transforms workflows.
Specialization beats generality - Domain-specific models (like Composer for code) can match frontier intelligence while optimizing for specific use cases.
Smart planning + fast execution - Hybrid architectures combine deep reasoning models for planning with specialized models for implementation, optimizing both quality and velocity.
Education bridges the gap - Many developers use AI without understanding fundamentals (determinism, context windows, token limits), creating opportunities for better education.
About Cursor
Cursor is an AI-powered code editor built on VS Code that integrates Claude and specialized models directly into the development workflow. The company launched Composer, their first proprietary frontier coding model built with reinforcement learning, focusing on speed and code generation efficiency.
Conference Appearance
Event: AI Engineering Code Summit 2025 Date: November 21, 2025 Time: 9:51 AM - 10:10 AM Session: Building a fast frontier model with RL
Lee presented on the infrastructure, training, and evaluations used to build Cursor Composer. He discussed the speed-intelligence tradeoff, RL-based training for code generation, and the practical architecture of using smart models for planning combined with fast models for execution.
Key Insights
“Composer is faster at similar intelligence. People really liked the speed.”
“Use smart models to make the plan, then use Composer to rip through the code.”