AI Product Engineer
I build your product. The whole thing, fast.
I design, build and ship complete products and automations for founders and engineering teams. One engineer, full accountability, from architecture through deployment.
Open work
Two things you can open right now.
No form, no call, no permission. Look at the product or read the source and decide for yourself.
Motorlane
Three dealerships, two currencies, one codebase.
A white-label showroom platform for car dealerships. Each tenant gets its own branding, currency, timezone and number-plate format out of a single deployment, with inventory, finance calculations and enquiry capture built in. Use the dealership switcher in the corner to watch the same codebase run a different market.
Three of the dealerships in the demo are invented, and that is the point: they exist to prove the tenancy and localisation layers hold across markets. An independent used-car dealership is running the platform as a live pilot.
dealer-seo-audit
Small enough to read in full, and worth reading.
That rule exists because of a real mistake. A false claim in front of a prospect costs the meeting, and the prospect's developer will always check.
234 lines, five functions, standard library only. It checks a dealership site for the failures that quietly cost enquiries: duplicate titles, missing share previews, absent structured data, soft 404s. Read the whole thing and judge the code yourself.
Selected work
The work I cannot show you.
Most of what I build is covered by NDA: enterprise engineering platforms, logistics operations, internal tooling. Here it is at the architecture level, led by the decision that shaped each one. Every figure is real and measured.
Engineered systems
Three systems I built end to end.
I engineer business-critical AI automation — a deterministic decision core, evidence grounding, human governance and controlled execution. Each one shown here is a showcase build: real engineering, mock data, dry-run only.
Distinct professional roles, each with its own purpose-built workflow
Build-failure diagnosis, per cycle
Manual failure-investigation effort eliminated
How I work
One engineer, end to end.
No project manager relaying messages to the person writing the code. One number and one scope before I start. Everything lives in your accounts from day one: code, credentials, infrastructure. Backend, frontend, architecture, deployment, with one person accountable for all of it.
30-minute call.
You tell me the problem: the manual task, the product idea, whatever's costing you time.
A fixed quote.
One number, one scope. No hourly meter, no surprise invoices.
A small first project.
Low-risk, fast-turnaround proof before you commit to anything bigger.
We scale from there.
Once you've seen the work, we take on the larger build.
FAQ
Questions clients ask
You're one person. What happens if you get hit by a bus?
Is my code and data secure?
What if the project doesn't go the way I expect?
Can you work on a project for one of our clients?
Will you sign an NDA? Who owns the IP?
What happens to what I send through this site?
Start here
Automation Audit
$500 fixed. Five working days.
You get written findings and one working prototype of the fix. You keep the code whether or not we work together again, and the $500 comes off anything that follows.
Run a studio or consultancy? The same $500 covers one defined ticket off your backlog instead: fixed scope, five working days, and the code is yours.
Show me the process eating the most manual time in your team's week.
Contact
Got something to build?
Tell me what it is. I reply within one business day, and you don't need a sales call to get an answer.
Book a call
Thirty minutes, straight to me. No assistant, no sales call, no deck.
- You tell me the problem, in your own words.
- I tell you straight whether I'm the right person for it.
- If I am, you leave with a scope and a fixed number.
Send an enquiry
A sentence or two about the problem is plenty. I reply to every message personally.