From sketch to production — built on purpose, not from a template.
I use APIs most teams never touch, build production sites on modern infrastructure, and ship the tools and systems operators can see but can’t build themselves.
Custom by default. Someone has a problem; I build what solves it.
Custom automations & tools
API integrations, SEO and AIO data tools, the unglamorous plumbing that holds up in production. Most of it is custom — someone has a problem, I build the solution.
Sites and web apps
Fast, modern websites and apps on Astro, Cloudflare, Vercel, Supabase or your infrastructure of choice. They load fast, they hold up, and the structure is decided on purpose — not stamped from a template.
AI that actually ships
Predictable pipelines with AI steps wired inside them, not a chat box with a nice interface. The hard part is the plumbing around the model — the error handling, the data, the parts that make it survive contact with real use.
Five takes. If the work doesn’t tie back to one of them, I don’t do it.
Real AI is engineering, not prompts.
Most “AI products” are a chat box with a nice interface. The money is in the plumbing underneath — the pipelines, the error handling, the data.
AI needs scaffolding, not freedom.
Real systems are predictable pipelines with AI steps inside them. Not bots making big decisions on their own.
The value is in the build, not the model.
Switching from one model to another changes maybe 10% of a working system. The other 90% is how it’s wired together.
Using AI isn’t the same as building with it.
Most “AI experts” are heavy users of AI tools. Shipping real AI-powered systems is a different skill.
Honest beats impressive.
If something’s a bad idea, or you don’t need it, or the boring fix is the right one — you’ll hear that first. I’d rather lose the work than ship you something that doesn’t hold up.
I was an operator long before I was anyone’s engineer.
Front-end builds
Joomla and Drupal, through WordPress, to modern static sites, complex web apps and tools. I was shipping production front-ends before half of today’s stack existed.
SEO, the operator side
Not advising from the outside — running it. I know what the levers actually do, because I’m the one who pulled them.
Python & AI
Real pipelines with real error handling, not prompt demos. The plumbing underneath, not the chat box on top.
Self-taught
Nobody handed me a framework. Pulled systems apart until they worked, then built what should have existed.
- Led teams through competitive, high-stakes niches
- Content automations for news publishers — AI and APIs at volume
- LinkedIn automation, end to end
Automation does the volume. The judgment calls stay human — that’s where it holds up.
Scope
Understand the problem, the data, and what access I need — and what should exist that doesn’t yet.
Research
Map the data sources and APIs in play, or your own systems, whatever the job actually calls for.
Architect
Design the pipeline: the structure, and where an AI step earns its place versus where it doesn’t.
Build
Wire the APIs, write the automation, stand up the app or website. The repeatable parts are already built.
Verify
Run it against real data, handle the errors, confirm it holds under real use. This is the quality gate.
Ship
Deploy and hand it over. Fast loading sites or web apps, infrastructure for the tools.
Got something that needs building?
Tell me what you’re actually trying to do — not the polished version. If it’s a fit, I’ll say so. If it isn’t, I’ll tell you that too. First conversation’s free.
Got it.
I’ll get back to you, usually within a day. No autoresponder, no sequence — just me.