Andrej Karpathy
Training & research
Makes model-training experiments concrete and accessible through compact public implementations.
The people behind the progress
Artificial intelligence moves fast. Follow the builders, explore their work, and keep the evidence close.
18 of 18 people
Training & research
Makes model-training experiments concrete and accessible through compact public implementations.
Translation & storytelling
An emerging builder combining visual understanding, dialogue context and comic translation in a local application.
Local models
Useful perspective on efficient models, speech and machine learning on personal hardware.
Experimental creativity
An emerging creative follow for agents that generate and iteratively inspect digital LEGO designs.
Sports audio & production tools
Builds practical tools for athletics and live production, including Stadium Sound, a soundboard for sporting events.
Image, video & audio
A useful anchor for visual model workflows, reusable examples and ecosystem developments.
Enterprise agents & workflow orchestration
Shares work on TransitionManager and coordinating agents in enterprise workflows at Transitional Data Services.
Work & adoption
Useful for tracking how model capabilities translate into work, interfaces and organizational choices.
Broadcast & live production
Connects AI assistants to practical editing and live production tools. Especially relevant to production workflows and Office Hours Global.
Local models
A specialized follow for model compression and local inference improvements.
Broadcast & live production
Production tools for bringing Zoom participants into live video workflows.
Broadcast & live production
Cloud production tools designed around production operators, with infrastructure for live media workflows.
Image, video & audio
Research-backed experiments in making visual generation more controllable and practical.
Coding agents
Worth following for agent architecture, coding tools and extensibility.
Image, video & audio
Useful for training and adapting creative models, rather than only consuming finished generators.
AI education
An especially useful follow for visual explanations that make technical model behavior understandable.
Training & research
Useful explanations of model architecture, training and the mechanics behind coding agents.
Practical AI
A strong source of hands-on experiments, inspectable tools, and practical analysis of language models.