DevOps & Deploy — Five Platforms, One Command

Deployment is where hobbies become products. I maintain deploy paths on six platforms and treat "one command" as a hard requirement — because when a platform turns hostile or a region goes down, migration speed is uptime.

Platform playbook

Platform	What I run there
Hugging Face Spaces	Hermes Stack, AI bridges — generous free compute.
Cloudflare Workers	usf-edge panels — KV state, 300+ PoPs, no cold starts.
Railway / Render / Fly.io / Koyeb	Panels and APIs — each with its own anti-blocking quirks.
VPS (Linux)	Classic deployments, tunnels, monitoring.

• Docker when the platform supports it; Procfiles and buildpacks when it does not.
• CI habits — GitHub Actions for tests and builds; deploys stay boring on purpose.
• Observability — structured logs, uptime pings from my own tooling, alerting to Telegram.
• Secrets discipline — environment variables, encrypted vaults, nothing in git.

The philosophy matches my networking mindset: every deploy is one takedown away from dying, so every deploy has a documented twin on another platform. My panels literally ship five deploy commands for this reason.