About
WumboLabs is my corner of the internet for local AI, Linux, homelabs, self-hosting, and building useful things.
My interest in local AI goes back to 2021, when I started experimenting with projects like DeepDaze. At the time, hardware limitations meant most experiments were small-scale, but the idea stuck with me.
What began as curiosity around local AI eventually expanded into Linux, self-hosting, Docker, storage, networking, automation, and homelab infrastructure. About a year ago I started building a more serious homelab environment and quickly discovered that every project led to three more.
Along the way I've migrated storage, rebuilt DNS infrastructure, broken Docker stacks, benchmarked local models, documented recovery procedures, and spent more late nights troubleshooting than I'd like to admit.
Today my primary focus is practical local AI.
I'm less interested in benchmark leaderboards and marketing claims, and more interested in answering a simpler question:
What can normal people actually run on consumer hardware?
That question led to projects like Monolith and LLMGauge, both built around evaluating local models in realistic conditions rather than idealized benchmarks.
The goal isn't to chase hype.
The goal is to figure out what is actually useful.
Professional Background
My technical work started professionally in support and network operations.
I worked in customer support, service assurance, and technical support roles troubleshooting wireless networks, hardware, configuration issues, outages, vendor coordination, field technician support, ticket documentation, and customer-facing communication.
That experience shaped how I approach systems:
- Isolate the failure point
- Document what changed
- Avoid guessing when evidence is missing
- Communicate clearly
- Keep rollback paths available
Before that, I worked in kitchens and eventually became a Sous Chef.
While the work was very different, many of the habits carried forward: stay organized, stay calm under pressure, solve problems quickly, and keep things moving when something breaks.
Current Areas of Interest
These days most of my time is spent somewhere between:
- Local AI and LLM evaluation
- Linux workstations
- Homelab infrastructure
- Docker and self-hosting
- ZFS storage
- DNS and networking
- Automation and tooling
- Documentation systems
How I Work
My general philosophy is simple:
- Stability before novelty
- Incremental changes over risky rewrites
- Documentation before memory
- Evidence before assumptions
- Clear rollback paths
- Practical tools over unnecessary complexity
Most of the things I build start with a simple thought:
"I wonder if this will work."
Sometimes it does.
Sometimes it fails spectacularly.
Either way, there's usually something worth learning.