# K-Dense-AI/karpathy

An agentic Machine Learning Engineer

Repository: https://github.com/K-Dense-AI/karpathy
Canonical: https://ross.abutalabs.com/products/karpathy
Homepage: https://k-dense.ai
Language: Python
License: MIT
License Family: permissive
Topics: agentic-ai, automl, machine-learning
Last push: 2026-08-18T23:06:42+00:00

## Health v2 (maintenance only)
Score: 60/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 98, release rhythm 35, longevity 20
- inputs: {"age_days": 290, "days_push": 15, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1532, forks 176 (observed 2026-08-28T04:04:59.428348+00:00)

## What it is
Karpathy is an open-source agentic Machine Learning Engineer that trains state-of-the-art ML models using the Claude Agent SDK and Google ADK. It serves as a reference implementation demonstrating K-Dense's Scientific Agent Skills library for machine learning workflows.

## Use cases
- automatically train machine learning models from a prompt
- use an AI agent as an ML engineer
- try agentic automl with Claude Agent SDK
- demonstrate scientific agent skills for ML tasks
- build an autonomous model-training pipeline
- run an AI agent that handles end-to-end ML workflows

## When to choose
- you want an agent that autonomously trains ML models
- you already use Claude Code and want to extend it with ML skills
- you want a simple reference implementation of Scientific Agent Skills
- you want to experiment with agentic automl in Python

## When to avoid
- you need a hosted, polished ML platform with support and compliance (use K-Dense Web instead)
- you want a traditional AutoML tool with a GUI
- you cannot use Claude Code or Google ADK
- you need production ML infrastructure rather than a demo agent

## Facets
- artifact type: application
- maturity: active
- function: agent-framework, machine-learning, llm-inference, developer-tools
- domain: machine-learning, artificial-intelligence
- platform: python, cli, cross-platform
- tags: agentic-ai, automl, claude-agent-sdk, google-adk, scientific-agent-skills, ml-engineer-agent, ai-agents, automation

## Member repositories
- K-Dense-AI/karpathy (main) score 60

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:59.428348+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:31:17.871221+00:00, confidence not recorded.
  - readme: https://github.com/K-Dense-AI/karpathy (fetched 2026-08-28T04:04:59.428348+00:00, sha 08c54be5d3ee)
  - homepage: https://k-dense.ai (fetched 2026-08-29T11:33:32.783277+00:00, sha 1ada73af7833)
  - site_page: https://www.k-dense.ai/about (fetched 2026-08-29T11:33:32.794573+00:00, sha d2ae8a64ebdc)
  - site_page: https://www.k-dense.ai/docs (fetched 2026-08-29T11:33:32.796297+00:00, sha b19d8a17bdaf)
  - site_page: https://www.k-dense.ai/pricing (fetched 2026-08-29T11:33:32.792508+00:00, sha 447642fdaf5e)
  - site_page: https://www.k-dense.ai/faq (fetched 2026-08-29T11:33:32.797876+00:00, sha c2c243ce74d7)
  - site_page: https://www.k-dense.ai/blog/introducing-waypoint-agent-skill (fetched 2026-08-29T11:33:32.799634+00:00, sha 15c385b655ba)
- Data as of 2026-08-30T08:39:29.467469+00:00.