Infrastructure & DevOps

Relying entirely on third-party platforms means giving up control over uptime, pricing, and data. The infrastructure behind this portfolio is deliberately self-hosted where it counts — a private Gitea instance for version control, dual Synology NAS units for automated backups, and Vultr VPS servers managed through CloudPanel for multi-site hosting. This setup provides full data sovereignty, predictable costs, and the flexibility to configure everything from nginx rules to deployment hooks without waiting on a support ticket.

The deployment workflow is designed around git push. Code goes from a local machine to the office NAS, where a post-receive hook triggers rsync to production servers automatically. A second push to the home NAS ensures every project is mirrored in a geographically separate location. Sixteen projects are tracked this way, each with its own bare repository, hook scripts, and deployment target — no CI/CD service subscription required.

The MCP (Model Context Protocol) servers add another layer to this infrastructure. Five custom Node.js servers expose AI models from Gemini, Groq, Ollama, GLM, and OpenCode as development tools that integrate directly into the coding workflow. These aren't wrappers around chat interfaces — they're structured tool servers that accept specific inputs and return formatted outputs, used daily for research, code review, content generation, and second-opinion reasoning across every project in this portfolio.