Ongoing · Infrastructure & AI
A self-managed home lab for services and local AI
To stay sharp and keep building, I run a self-managed virtualization lab: XCP-ng hosting Debian VMs for PostgreSQL, Nginx, and Node.js APIs, plus local LLM and image-generation workflows I use in day-to-day development.
Timeframe
Nov 2025 – present
Hypervisor
XCP-ng
Services
PostgreSQL, Nginx, Node.js
AI
Local LLM + Stable Diffusion
Problem
A real environment, not a tutorial sandbox
I wanted a place to keep operational fluency sharp and try modern tooling end to end. Running actual services on hardware I manage forces the same decisions a production system does: networking, persistence, process management, and recovery.
Approach
Virtualized services on hardware I own
The lab runs XCP-ng as the hypervisor with Debian virtual machines for multi-service hosting. I operate PostgreSQL, Nginx, and Node.js API services in it, which keeps my hands on the database, reverse-proxy, and deployment layers I use professionally.
On top of that, I set up local AI workflows: a local large language model (Qwen via oMLX) and Stable Diffusion through ComfyUI for image generation. Day-to-day coding happens in Zed, which is how I have been shipping recent work quickly.
Results
What it gives me
- A realistic environment for standing up and operating Node.js, PostgreSQL, and Nginx.
- A private staging ground for side projects before they touch the public internet.
- Hands-on familiarity with local LLM and image-generation tooling and AI-assisted development.
Lessons
Why it matters for hiring
This lab is evidence that I keep operating real systems, adopt current tooling, and can still reason about infrastructure from the metal up — the same mindset I bring to shipping and maintaining production software.