I’ve been experimenting with automating the setup around model training — choosing compute, provisioning GPUs, and preparing the training environment

Hey everyone,

I’ve been working on a tool called Orchard Compute.

The idea is simple: connect it through MCP, then describe what you want to do in one sentence.

For example:

“Fine-tune this model on my dataset with the best GPU for my budget.”

It can handle things like choosing/provisioning compute and setting up the training environment, so you don’t have to manually compare GPU providers or configure everything each time.

I’m looking for people who already train or fine-tune models and would be willing to try it and tell me where it breaks.