Homelab platform: how and why

High-level notes on a personal platform run with production practices: real Kubernetes, infrastructure as code, GitOps and pipelines with no static credentials.

This site shows patterns, decisions with trade-offs and lessons learned. The technical detail of each component (configuration, endpoints, runbooks, roadmap) lives next to the code, in private repositories.

In one picture

flowchart TB
    dev["Developer"] -- "PR + merge" --> git["Git repositories"]
    git -- "CI: image build" --> reg["Image registry"]
    git -- "desired state" --> argo["GitOps controller"]
    argo -- "reconciles" --> k8s["Kubernetes cluster"]
    reg -. "pull" .-> k8s
    cloud["Cloud: identity and secrets"] -. "OIDC / secrets" .-> k8s
    user["User"] --> edge["Edge (tunnel, no open ports)"] --> k8s

Where to start

• Architecture: the layers and how they relate.
• Decisions: what was chosen, what was left out and what each choice costs.
• Lessons learned: what broke or surprised.
• Case studies: real applications, described abstractly.

Principles

• Everything declarative. Git is the only source of desired state.
• No long-lived credentials. Federated identity in CI and secrets injected at runtime.
• Convention over configuration. A repository's name already says what it does.
• Docs next to the code. Each repository documents itself; this site summarizes what is worth sharing.