I like running my own stuff. Over the years that turned into a proper little lab, spread between my home network and a rented server in the cloud.

The hardware

• Several servers at home, mostly running Proxmox. One is the production box that the always-on things live on. The others are for experiments, spare capacity and breaking things safely. Not all of them are switched on at any given moment.
• A rented Linux server in the cloud for the things that need to be reachable from anywhere.
• My MacBook, which is also where I run local language models and do most of my research.
• A shelf of old gear that I like to bring back to life. See Second Life.

What runs on it

Quite a few apps and ready-made services at the moment, spread across a couple of servers. The kinds of things:

• My own projects, including this website, the Inference Gateway and a few web apps and APIs.
• Self-hosted tools I’d otherwise pay for: uptime monitoring, workflow automation, document signing, one-time secret sharing, an API client, a chat interface for local models and a WordPress management hub.
• A media server and its helpers.
• Test and demo environments, because I’d rather break my own stuff than someone else’s.

Holding it together

Shipping is mostly a git push, and anything public sits behind proper access control. A lot of it only exists on my private network, and that’s on purpose.

I don’t document the details here, but I’m always up for talking about self-hosting. If you’re setting something up and want a second opinion, ask.