Agent-first ML training management for your home lab

BeekeeperML turns your home lab server into an agent-ready ML training platform. Install it once, then let your AI agents handle the rest.

Download View Docs
BeekeeperML dashboard showing a live training run with log output

Built for the way you actually work

Agent-first MCP API

Connect Claude, GPT-4, or any MCP-compatible agent directly to your training server. Start runs, check status, pull logs — all via tool calls.

Parallel training runs

Run multiple branches simultaneously. Each parallel run gets its own workspace, log file, and TensorBoard instance — ideal for A/B experiments.

TensorBoard built-in

Metrics embedded in the UI, no separate setup required. BeekeeperML starts and stops TensorBoard automatically with each training run.

Runs on your hardware

Install on any Linux server with a GPU. Your data, your compute, your control. No cloud dependency required.

Free forever for home lab users.