Selfhosted

https://docspell.org/ for organizing your documents using machine learning.

Maybe not a service in the typical sense, but setting up your router+server to route your home network traffic through a VPN is a fun project.

My router (MikroTik) supports WireGuard, so I can use it with Mullvad for the whole house---but wg is demanding and it's a slow router, so while it can NAT at ~1Gbps, it can't do WireGuard at more than ~90Mbps. So, I set up WireGuard/Mullvad on a little SBC with a fast processor, and have my router use that instead. Using policy based routing and/or mangling, I can have different VLANs/subnets/individual hosts selectively routed through the VPN.

It's a fun exercise, not sure I implemented it in a smart way, but it works :)

• For LLM hosting, ik_llama.cpp. You can really gigantic models at acceptable speeds with its hybrid CPU/GPU focus, at higher quality/speed than mainline llama.cpp, and it has several built in UIs.

• LanguageTool, for self run grammar/spelling/style checking.

I started with NextCloud, mainly so I can start synchronizing Joplin notes. Maybe I could hook it up to also sync Logseq?

I chose this VTT because it's dead simple and description on owlbear legacy did not sound encouraging

Then, on my list I have

• LibreTranslate
• tootik

If you have a Nvidia graphics card 1070 and above, then openwebui. You can selfhost your own LLM. AMD is probably supported but haven't checked.

RSSHub. Being able to get all my updates in one place changed how I view the internet for the better.

Adguard Home, with domain pointed to it and using it as Private DNS on Android. No more ads anywhere!