Self-hosted AI, documented honestly

I run large language models on my own hardware (NVIDIA DGX Spark) and write down what actually breaks, and how I fix it. No cloud, no vendor lock-in, no marketing gloss.

• 📓  Blog  ·  sovgrid.org    140+ engineering logs on running AI at home
• 🟣  Nostr  ·  cipherfox@sovgrid.org
• 📡  RSS  ·  sovgrid.org/rss.xml
• ⚡  Value-for-value  ·  cipherfox@rizful.com

_{Topics: SGLang · vLLM · quantization (NVFP4 / FP8) · OpenWebUI · Caddy · Docker hardening · backups · the boring failures nobody blogs about.}