EDGEGRID

A decentralized grid for training ML models on GPUs people already own

Submit training jobs to a pool of volunteer GPU workers, or contribute your own idle GPU to the grid. No cloud bill, no cluster to manage.

SUBMIT JOBS

Push a training script, dataset ref, and hardware requirements. The grid places your job on a matching worker automatically.

CONTRIBUTE GPU

Join as a worker node with a single command. Approve or reject jobs before they run on your hardware.

LIVE MONITORING

Stream logs in real time, track job state, and download checkpoints the moment training finishes.

HOW IT WORKS
  1. 01Sign in with GitHub
  2. 02Submit a training job or join as a worker node
  3. 03The coordinator matches jobs to available hardware over Raft-backed consensus
  4. 04Watch logs stream live, pull the checkpoint when training completes
EDGEGRID — DISTRIBUTED ML TRAINING