About

What this is #

The Scaffold Rack is a hands-on platform for building and operating a modern homelab — documented publicly as I go. It’s a ground-up rebuild of my homelab into something closer to real platform engineering, and every meaningful decision, reversal, and dead end gets written up.

Scaffolding is temporary and iterative by nature. The name is a signal: this is building in public, not a finished reference architecture. If a decision here turns out to be wrong three months later, the post gets a callout at the top rather than a silent edit. The value is in the honest record, not the polished outcome.

Who I am #

I’m Andrew Krull. I’ve been doing open source systems work for 20+ years — sysadmin, then DevOps, now platform engineering. Currently a senior engineer focused on internal platform tooling: GitOps pipelines, observability stacks, and the kind of infrastructure that other engineers actually want to use.

Self-hosting has been part of my life for a long time. What’s changing here is the rigor — I’m operating my home infrastructure like production: declarative, version-controlled, observable, recoverable. Not because my home needs five-nines uptime, but because the discipline is what makes interesting architectures possible. A homelab that can be rebuilt from git is a homelab you can experiment on without fear.

Why I’m writing this #

Three reasons, in decreasing order of selfishness:

1. I learn by writing. Explaining why something works forces me to actually understand it. Half of what I know I learned by writing it down.

2. There’s a gap in the public material. Plenty of blogs show you the finished Kubernetes cluster or the clever Terraform trick. Fewer show the month of “why is MetalLB doing that” that came before it. I want to fill some of that.

3. The AI platform story is underdocumented. I run local inference at home, and I’m building toward a self-hosted AI platform with governance, observability, and real tooling integration. That’s increasingly relevant work and almost nobody outside big companies is writing about it.

What you’ll find here #

• Series covering multi-post arcs: the initial build, the AI platform work, specific deep-dives on subsystems that warranted more than one post.
• Standalone posts on decisions, tradeoffs, things that broke.
• Referenced commits. Posts that change code link to specific SHAs so you can see exactly what was done. Nothing is hand-waved.
• The occasional tangent. Some of this is technical memoir. I like that.

Code is Apache 2.0. Writing is CC-BY-4.0. Use it, build on it, send corrections.

Reaching me #

Comments are off. They’re a maintenance burden I don’t want and they tend to attract the wrong kind of engagement for technical writing.

If something in a post is wrong, unclear, or missing context, open an issue on the relevant repo at github.com/scaffoldrack. Issues are better than comments for this kind of material: they’re public, they’re tied to specific artifacts, and they don’t disappear when the comment-hosting service shuts down.

For anything else, you can find me on GitHub.