One engineer,
the whole system.
Rabbit Hole Digital is a tiny studio, on purpose. When you hire us, you work directly with the person who does the discovery, writes the code, and stays on after launch. No account managers, no handoff to juniors.
I have spent a career making complicated things behave.
I come from a mixed background, and I have come to see that as an advantage rather than a detour. Making sense of many technical disciplines at once, and finding the thread that connects them, is exactly what good software and automation work asks for. It matters even more with AI, where most of the value sits in the joins between one piece of technology and another.
That range for me, comes from years in the music industry. I have worked as a an audio engineer, a sound designer, educator, and a musician, and along the way I touched most of the rest of it too: marketing, sales, software development, product management. Different rooms, same instinct for how the parts fit together.
I started out writing software and building solutions. Reliability first. Clarity over cleverness. A bias toward the smallest system that solves the real problem. I moved from building products to untangling them: joining the systems that had grown up in isolation, automating the work nobody should be doing by hand, and more recently, putting AI to work where it genuinely earns its place. Rabbit Hole Digital is where all of that comes together.
The name is a nod to how the good problems go. You start with a simple question, and the real work is everything you find on the way down. I like it down there.
What I actually do.
- LLM application design
- RAG & retrieval
- Agents & tool use
- Evals & guardrails
- Prompt engineering
- On-device & local models
- TypeScript & Python
- React & Next.js
- WordPress & PHP
- APIs & data modelling
- Postgres & Supabase
- Testing & observability
- Workflow orchestration
- Integrations & webhooks
- Discovery & scoping
- Technical writing
- Stakeholder translation
- Long-term ops & support
Three principles I do not bend.
Build less.
The smallest system that solves the real problem is almost always the right one. Restraint is a feature.
Reliability first.
Boring, well-tested software you can still understand in a year beats clever software you cannot.
Be honest about AI.
Models where they add value, rules where they do not. I will talk you out of the ones that add risk.
You get the rare thing here: someone senior enough to tell you what not to build, who then goes and builds the rest properly.