Senior Product Designer · Auckland
Most of design is decisions — what to make, why now, whether it worked. Ten+ years across financial and technical products taught me to trust mine.
Selected work
Open-source AI tooling at Galvanized Pukeko: a CLI-based code-review assistant and a UI-generating framework. Conversational-AI interface design applied in the open — where the work and the constraints are visible to anyone who wants to look.
1,000+
developers using the Gaunt Sloth CLI assistant. Open-source, shipped, in production.
AI Product Contributor · 2024 → present.
Two sides of one business model: a 2026 design challenge, built end to end.
A functional prototype, hand-coded in Claude Code & Zed — not Figma.
58%
lift in customer satisfaction after the order-placement redesign
93%
drop in users complaining the product was untruthful
Mobile app + 4 wearable sensors for post-surgery rehab. The full case: a screen-by-screen walkthrough of the surgeon-to-physio pivot, then the shipped data-visualisation screens that made four sensors readable in seconds.
15%
lift in trial-to-customer conversion after the IMU Step iOS re-launch — what the customer voices below were responding to
Senior Product Designer · IMeasureU / Vicon
From the field
The data collection is robust, and I love the way the info is organised in the graphs.
The new phone app is great. It works fantastic. Really like being able to footnote a group simultaneously.
Knee Monitor added value to me as a PT and my clinical practice — another tool I can use with my patients, and it's very visual. I can easily observe progress and compare between sessions, and patients get the benefit of seeing the changes in the report. Easy to comprehend, and for patients to track their own progress.
Patient report and the graphs are visually appealing and easy to understand. I like being able to click a particular time point and have it highlight the results from that session across the different tests. It's also appealing to the patients.
The visuals are very nicely done on the app, and my patients really liked seeing the visual feedback.
This is exactly what I want to see.
Also in the field with the Buffalo Bills and the Portland Trail Blazers, who moved to IMU Step for faster set-up, better battery life and clearer dashboards.
Recognised earlier, too: Great User Experience Award, 2017 — FinancesOnline, for the 9 Spokes fintech dashboard.
How I work
The cost of producing UI has collapsed. What used to take a week of Figma now takes an afternoon in Cursor with Claude Code. That changes where the value sits — less in production, more in deciding what's worth producing in the first place. So that's where I spend my time: framing the problem, choosing the direction, then building the prototype myself rather than describing it.
In practice that looks like this: I start with the outcome and the constraints, sketch the shape of the solution, then move into a coded prototype early — usually Vue or React inside an existing design system, deployed to a branch URL that stakeholders can actually use. That's how the day-to-day work at Defrr and Galvanized Pukeko gets shipped. The property-platform design challenge above — buyer and seller side — was built the same way, end to end.
I've spent the last year deliberately closing the technical gap that used to sit between designers and engineers. IBM's AI Product Manager Professional Certificate and four Generative AI courses through IBM SkillsBuild covered the model and product fundamentals; Google Cloud's Conversational AI on Vertex AI and Dialogflow CX gave me hands-on experience designing for agent behaviour; LinkedIn Learning's AI Product Ideation track filled in the discovery side. Combined with shipping production code at Defrr and Galvanized Pukeko, it means I can discuss trade-offs with engineers as a peer, not hand off a static file and wait.
The constant across all of it is taste. AI gives you a starting point. Whether it ships as something coherent or something generic — that's still a human call.
Working with me
“The board says we need AI in the product. By Q3. Figure it out.”
That brief is landing on product teams everywhere — and most don't have anyone who has actually built with a model, wired up an agent, or shipped a tool call to production. What they need isn't another vendor deck or another chat box bolted onto search. It's someone who can hold the trade-offs under sceptical questioning, and who understands both the AI architecture and the product surface.
The property platform above is the worked example. Not a chat box pasted on the side — a product redesigned around what AI is actually good at: the suggests, never decides posture, the router-and-specialist architecture, sourced data only, and the agent-economics trade-off named honestly rather than hidden.
If your board has handed you a brief that looks like that — this is the kind of conversation I have for a living.
Background
Conferences: ConFig 2023 / 2024 / 2025 (online), Design Buddies (May 2024), 10xDesigners (Feb 2024), UX New Zealand (Wellington, Feb 2023).
Full CV (PDF) →Contact