Welcome to the My Home Lab repository. This project documents my home lab, a personal infrastructure built around data privacy and digital autonomy. Rather than handing control of my data to third-party platforms, I self-host the services I rely on daily, keeping everything under my own roof. It's a living system that grows as I do, and this site exists as both a reference for myself and a transparent look at how it all fits together.
The home lab has been reformed into a unified, privacy-focused architecture built around a single high-performance computer:
- Services run on a dedicated compute system
- Access is provided securely through Tailscale mesh networking
- Control plane is self-hosted via Headscale on my VPS
- Authorization is managed by ACRO (similar to Authentik, with deeper code-level control)
- Domain protection via my domain with ACRO-secured access
This approach provides granular control, privacy, and security while maintaining ease and reliability.
ECHO Music Streaming Currently, the main service running on this infrastructure is a self-hosted music streaming platform built around Navidrome. I developed a custom frontend called ECHO to tailor the experience to my specific needs.
ECHO integrates directly with ACRO for Single Sign-On (SSO), ensuring that only authorized users can securely authenticate and access streaming content from my personal music library. This setup demonstrates a complete full-stack deployment: managing the infrastructure, securing the network, and developing both the user-facing application and its authentication layer.
Gitea Source Control & CI/CD Another core service is my self-hosted Gitea instance, which acts as the central hub for my coding and game development work. It heavily utilizes Git LFS (Large File Storage) to reliably handle the large binary assets required for my Unity projects. Coupled with a local runner for CI/CD tasks, this provides a completely private, high-performance development pipeline.
Note: This diagram shows the original MK-1 setup and is no longer active.
The original setup featured multiple distributed nodes for learning purposes:
ENLIL- RETIRED - Local DNS resolver with Pi-hole- CASPER - NAS storage (still active)
LORIC- RETIRED - Orchestrator node and CI/CD runnerAUREL- RETIRED - Worker node for heavy builds- LABRYS - Storage device
MK-1 was valuable for understanding distributed CI/CD infrastructure, but has been consolidated into the cleaner MK-2 design.
- The setup above will need to have their static ips migrated to a much lower ipv4 range, this is so I can tell the router to avoid assigning any future devices on the network to be accidentally assigned to one of these critical ips. There are many ways to go about this so for the moment it will be on my TODO.