Tag
The author describes building a self-improving 'company brain' using Claude Code and a wiki of markdown files, inspired by Andrej Karpathy, to organize and query training materials efficiently.
Andrej Karpathy shares his straightforward approach to working with AI, emphasizing plain language over complex prompt engineering, and predicts that by 2026 engineers who dismiss LLMs will lose to juniors who use them effectively.
Andrej Karpathy open-sourced an autonomous research agent that runs its own ML experiments overnight using a single GPU, automatically iterating on improvements by editing code and keeping changes that lower validation loss.