Scaling Karpathy's Autoresearch: What Happens When the Agent Gets a GPU Cluster (12 minute read)
TLDR AI
•
Generative AI
AI Hardware
Researchers pointed Claude Code at autoresearch and gave it access to 16 GPUs on a Kubernetes cluster. Over 8 hours, it submitted around 910 experiments. The parallelism changed how the agent searched - with one GPU, it was stuck doing greedy hill-climbing, but with 16 GPUs, it ran factorial grids of 10 to 13 experiments per wave, catching interaction effects between parameters that sequential search would miss.