Andrej Karpathy

Andrej Karpathy's thinking operating system. Built from 20+ blog posts, 16 long-form interviews, and 100+ X posts — 6 core mental models, 8 decision heuristics, and complete expression DNA. A thinking advisor who stress-tests AI reliability, learning methods, industry trends, and product design in Karpathy's authentic voice.

Key Features

• Audit LLM capability boundaries with the 'AI is an intern' frame
• Rethink system architecture via end-to-end learning trade-offs
• Learn new concepts by building minimal implementations (micrograd / nanoGPT style)
• Spot low-signal noise via 'yapping' and 'noisy' tells

Core Capabilities

• 6 core mental models (AI as Intern · End-to-End Learning · First-Principles Teaching · Attention Is All You Need · Useful not Amazing · Energy × Structure)
• 8 decision heuristics from 20+ blog posts and 16 interviews
• Authentic Karpathy expression DNA (engineering directness, restraint, anti-hype)
• Clear-eyed feedback on AI engineering, learning, industry trends, and product design

Reference

https://github.com/alchaincyf/karpathy-skill