← All selected work

Project case study · Systems infrastructure

A home lab operated like a small platform.

My home infrastructure is where I practice the unglamorous parts of engineering: identity, recovery, observability, scoped access, and keeping a growing service graph understandable.

Platform
TrueNAS + containers
Ingress
Cloudflare tunnels
Identity
OAuth/OIDC proxies
Scale
15+ routed services
Users + devicesweb · mobile · voice
Encrypted edgeDNS · tunnels · access policy
Homeautomation + voice
Mediaphotos + library
Knowledgenotes + recipes
AgentsMCP + AI gateways
TrueNAS foundationstorage · backups · containers · monitoring

01 · Operating model

Every new service must earn its place.

I treat deployment as a lifecycle: choose a clear owner, define ingress and identity, pin persistent data, add a health signal, document recovery, and decide what an automation agent is allowed to touch. That discipline matters more as the lab grows.

02 · Selected systems

Infrastructure assembled around real household workflows.

Voice

OmniRoute

A stateless Wyoming bridge routes Home Assistant audio to a compatible transcription gateway and local conversation agent.

Identity

One guarded edge

OAuth proxies and access policies keep public ingress separate from internal service credentials.

Automation

Home Assistant

Lighting, displays, recurring tasks, and voice flows are maintained as observable systems rather than isolated automations.

Operations

Recovery first

Persistent volumes, configuration backups, health checks, and scoped administrative tooling reduce fragile hand work.

03 · Takeaway

Reliability is mostly restraint.

The lab has taught me to favor small interfaces, reversible changes, explicit permissions, and boring recovery paths. Those habits carry directly into embedded and verification work: make state visible, constrain what can happen, and leave evidence behind.

Explore related code ↗