Hi HN, I’m Tao. I built ML Patron for a problem I kept running into with ML ideas: getting from “this is worth trying” to “the experiment actually ran and the result is inspectable.”

To me, there are usually three missing pieces that kill an idea: funding, execution, and continuity. Good ideas often stall because nobody wants to pay for that first GPU run, there isn’t a simple execution layer for reproducible runs with tracked metrics, and the surrounding context, like notes, discussions, and iteration history, ends up scattered across repos, chats, and docs.

ML Patron is my attempt to bring those pieces together. You can propose ML experiments, discuss them, fund them, and run them with a dry run first to catch bugs before spending the budget. If the dry run looks good and the run gets funded, the full experiment runs in a reproducible environment and all outputs are tracked.

I also built agent support in from the start. There’s a public skill.md and an API flow, so coding agents can use the platform directly instead of only acting through a human.

It’s been working well for my own research workflow, but I haven’t had real external users yet. I’m not sure which parts will generalize and which parts are too specific to how I work. I’d love to see how it fits into other people’s workflows.

If you have a non-trivial run in mind, please try it out. No need to pay for it yet. I’m happy to sponsor the runs for now while I figure out the rough edges.