Senior UX Architect & UX Engineer
I design and build enterprise telemetry, IoT hardware interfaces, and the full-stack applications that sit on top of them. Solder to type.
Domain channels
Bars are years active in that domain, roles included — not proficiency. The number counts shipped systems. Hover to filter deployments.
Stack coverage
Systems out of 9 where I owned that layer — a count, not a competence score.
Selected work
Taken from nothing to shipping.
Poultry telemetry on custom ESP32 hardware — sensor head to manager's lock screen.
Founder · firmware · product design2025 —
An offline-first app with a local vision model that reads handwritten ledgers into structured records.
2026 —
A social network scoped to one room, that forgets you when you leave it.
A D2C storefront built around a 30-minute freshness promise, from cold chain to checkout.
Natural MeatFounder & Product2025
Systems people use to do their job, built inside the organisations that run them.
Designed and built the internal purchasing platform ~600 independently-owned local companies buy through — and themed its component library.
WinsupplyUX Designer · design + front-end2022 — 2025
The student portal's shell was modernised; registration, records and payroll behind it were not. An audit of the live screens, and the system to replace them.
SCADUX Designer · research → design system2019 — 2022
Hardware and interface work I took end to end.
A Far-UVC disinfection system for rideshare cabins — enclosure, firmware and safety interlocks.
Enclosure + firmware + app2020
A carryable mesh radio and app for the hours after a disaster takes the cell network with it.
SCADProduct lead → product designer
A conversational take on the board game — an interaction experiment, not client work.
Interaction experiment2019
How I work
Each stage names what it produces and links to a project where you can open that artifact and check it. None of this is a claim you have to take on trust.
I go where the work happens and watch it — sheds, warehouses, counters. What people actually do becomes the constraint set, before a screen is drawn.
Build the risky part first and put it in front of people. Testing is what removes features here, not what confirms them — the list gets shorter before anything gets committed to.
Tokens, component contracts, and the rules that stop them forking — written as code the build reads, not as a palette sitting in a design file.
I implement it, ship it reporting its own state, and stay on it long enough to find out what I got wrong. Designed and built by the same person, so the seams land where they were meant to.