100% I love self-hosting all my services, self-hosting means freedom. I'm using old parts from my PC, email, nextcloud, tor node, game servers, and, yes, even AI stuff with the gpu I threw in there.

awesome! how do you do email delivery? i still have to route things through an external service to get that to really work. oh yeah, i gave my son access and he hosts Rust and Minecraft servers.

i'm looking into doing security cam + AI. just canceled my Nest subscription.

I self-host everything I can. Unfortunately email is a bit tough these days but I should start archiving it locally. What tools are you using for that?

I haven't done much experimenting with local AI models yet - probably need some newer hardware for that to be worthwhile.

I wrote some code with an IMAP lib. Wasn't too happy with out of the box things.

The smaller Deepseek models are most interesting for self hosting LLM's at the moment. Impressive! Many other ML Things need a lot less compute, though. Like facial recognition, spam filtering, etc.

How have you done the last one? I have stuff running on ollama but not with "access to all of it"