---
title: "Scaling Karpathy's Autoresearch: What Happens When the Agent Gets a GPU Cluster"
url: https://daily.dev/posts/scaling-karpathy-s-autoresearch-what-happens-when-the-agent-gets-a-gpu-cluster-hccbqaaty
source_url: https://blog.skypilot.co/scaling-autoresearch/
type: article
source: "SkyPilot"
published: 2026-03-19T05:25:47.375Z
updated: 2026-03-19T05:26:19.440Z
tags: ["kubernetes", "gpu", "ai-agents"]
reading_time: 12
upvotes: 26
comments: 2
language: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Scaling Karpathy's Autoresearch: What Happens When the Agent Gets a GPU Cluster

**[SkyPilot](https://daily.dev/sources/skypilot)** · 12 min read · 26 upvotes · 2 comments

## Summary

Karpathy's autoresearch project lets a coding agent autonomously improve a neural network training script by running experiments in a loop. This post scales that setup by giving Claude Code access to 16 GPUs (H100s and H200s) on a Kubernetes cluster via SkyPilot. Over 8 hours, the agent ran ~910 experiments in parallel waves of 10-13, achieving a 9x throughput increase over single-GPU sequential search. Key findings: parallelism enabled factorial grid search instead of greedy hill-climbing, allowing the agent to discover that scaling model width (aspect ratio 96) outperformed all hyperparameter tuning combined. The agent also autonomously developed a two-tier hardware strategy — screening ideas on cheaper H100s and validating winners on H200s — without being prompted. Total cost was under $300 in GPU compute plus ~$9 in Claude API fees. The full setup is available as an open-source example in the SkyPilot repo.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://blog.skypilot.co/scaling-autoresearch/>

## Community discussion

Top comments from developers on daily.dev.

**@tilman151** · 1 upvotes

> How did they arrive at 9$ in Claude API fees for their session? I ran the original autoresearch repo adapted to my personal machine and it chewed through 27$ of API costs in eight hours.

**@sanjeevbhandari01** · 0 upvotes

> Nice read

## Similar posts on daily.dev

- [Autoscaling Autoresearch: Give your agents elastic GPUs on Modal](https://daily.dev/posts/autoscaling-autoresearch-give-your-agents-elastic-gpus-on-modal-8y74qluv0) · Modal · 1 upvotes · 0 comments
- [Running Karpathy's autoresearch on Red Hat OpenShift AI: 198 experiments, zero intervention](https://daily.dev/posts/running-karpathy-s-autoresearch-on-red-hat-openshift-ai-198-experiments-zero-intervention-uejxfvugu) · Red Hat Developer · 1 upvotes · 0 comments
- [Andrej Karpathy’s 630-line Python script ran 50 experiments overnight without any human input](https://daily.dev/posts/andrej-karpathy-s-630-line-python-script-ran-50-experiments-overnight-without-any-human-input-pjqwmdwbv) · The New Stack · 45 upvotes · 0 comments

---

Tags: [#kubernetes](https://daily.dev/tags/kubernetes), [#gpu](https://daily.dev/tags/gpu), [#ai-agents](https://daily.dev/tags/ai-agents)

[View this post on daily.dev](https://daily.dev/posts/scaling-karpathy-s-autoresearch-what-happens-when-the-agent-gets-a-gpu-cluster-hccbqaaty)