# ml-developer

ml-developer — a dedicated subagent for training machine learning models, exploring hyperparameter spaces, and editing Jupyter notebooks.

- subagent · `type=agent` · Claude Code, Codex, Cursor · ★ 850 repo stars
- Author: Claude Code
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/ruvnet/agentic-flow/tree/HEAD/agentic-flow/.claude/agents/data/data-ml-model.md
- Updated: 2026-10-26
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/agents/ml-developer

## What it does

This subagent manages end-to-end machine learning workflows directly inside code repositories. It reads and edits files within directories like `data/`, `models/`, and `src/ml/`, executing training scripts and updating Jupyter notebooks. Through `claude-flow@alpha`, it caches experiment metrics, compares parameter configurations, and records successful optimization runs.

Its toolchain remains intentionally restricted. The agent cannot search the public web or run open-ended planning tasks, forcing it to rely exclusively on local training data. Datasets are capped at 100 MB, execution runs time out at 1800 seconds, and process memory cannot exceed 2 GB.

Deploying a model to production, launching large training jobs, or removing dataset files strictly requires explicit manual user approval.

## Install

### Claude Code (native platform)

**Manually: put the files in place**
- File: `.claude/agents/data-ml-model.md` — For all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention

### Codex

**Manually: put the files in place**
- File: `.codex/agents/data-ml-model.toml` — Codex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/

### Cursor

**Manually: put the files in place**
- File: `.cursor/agents/data-ml-model.md` — Cursor also reads .claude/agents/

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/ruvnet/agentic-flow/tree/HEAD/agentic-flow/.claude/agents/data/data-ml-model.md

## When to use it
- train classification and regression pipelines on local datasets
- tune model hyperparameters while logging optimization history
- edit and execute workflows in Jupyter notebooks and Python scripts
- hand off data preparation routines to a data-etl subagent

## Example prompts
- Build a churn classification pipeline using data/churn.csv
- Tune hyperparameters for the estimator configured in src/ml/train.py
- Run the training procedure inside notebooks/forecast.ipynb

## Similar
- [developer-kit-ai](https://affpapa.org/en/nevibecode/plugins/developer-kit-ai.md) — plugin · ★ 372 repo stars
- [experiment-pipeline](https://affpapa.org/en/nevibecode/skills/experiment-pipeline.md) — skill · 405 installs
- [optim-agent](https://affpapa.org/en/nevibecode/skills/optim-agent.md) — skill · ★ 796 repo stars
- [jupyter-notebook](https://affpapa.org/en/nevibecode/skills/jupyter-notebook.md) — skill · 6100 installs
- [adaptive-wfo-epoch](https://affpapa.org/en/nevibecode/skills/adaptive-wfo-epoch.md) — skill · 146 installs
- [jupyter-to-marimo](https://affpapa.org/en/nevibecode/skills/jupyter-to-marimo.md) — skill · 3600 installs

From the same set: [ruvnet/agentic-flow](https://affpapa.org/en/nevibecode/by/ruvnet/agentic-flow.md) — 76 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
