All articles tagged "tutorial" : self-hosted AI fixes, setups, and architecture notes.
Backing up model weights is the wrong abstraction. Backing up the model identifier, the configuration, the customer data, and the runbook is the right one. The weights are reproducible; the data and the runbook are not.
Read article →
Static site generation, no database, no PHP, no CMS. The blog is markdown files in a Git repository, built locally, rsynced to a small VPS, served by Caddy. Faster than any WordPress, immune to the WordPress vulnerability class, and the archive is plain text on disk.
Caddy on a small VPS, Cloudflare Tunnel to the home Spark, no inbound ports open at the residential ISP. Six hundred milliseconds of TLS handshake handled by a CDN that can absorb the abuse the operator cannot. The honest rented dimension and the migration path if it has to come back.
The hardware wallet holds the seed; the node holds the channel keys. The integration pattern is straightforward in principle and finicky in detail. Here is the working setup with the BitBox and the LND backend, plus the three integration mistakes I made before it worked.
Lightning is the sovereign payment surface. Running your own node is the difference between accepting Lightning and being Lightning. Here is the working setup, the operational discipline, and the failure modes that bit me in the first two months.
A Tor hidden service in front of a sovereign-AI endpoint is the right answer for three specific reader populations and the wrong answer for everyone else. Here is how to tell which population you are in, and the configuration if you are.
File-integrity monitoring is unnecessary on most workstations and necessary on a few specific ones. The AI box that runs other people's inference is squarely in the necessary category. Here is how to wire AIDE and Tripwire without producing alert fatigue.
A self-hosted Gitea instance holds the prompts, the unit files, the runbooks, the customer data references, and the model identifiers for the sovgrid AI stack. The pattern is mundane and load-bearing.
A step-by-step runbook for getting a DGX Spark back to full production after a power event. Thirty minutes if you have rehearsed; two to six hours if you have not. The procedure assumes a UPS for graceful shutdown and a separate management host.
Prometheus plus Grafana plus one phone number plus the discipline to never alert on something that is not actionable. The observability stack that lets one operator sleep through the night and still catch the failures that matter.
Six unit-file patterns that make a multi-service AI stack survive crashes, reboots, and power events without operator intervention. The patterns are not novel; the discipline of applying them consistently is.