The screens that decide whether a platform works are usually the last ones to get design attention. Configuration flows, investigation tooling, tables holding thousands of rows.
That gap is expensive. An unclear status becomes a support ticket. An ambiguous default becomes a payout error. It is also where renewal risk sits, because customers judge a platform on the tools their own team opens every morning.
So I work where attention is underpriced. Most of my week goes on the parts that never appear in a portfolio: reading the data model, arguing about field order, deciding which settings a person should never have to see. The visual work is the last ten percent.
I did not plan this. I started in consumer work, launching car rental inside a super app and leading a mobile suite, then took on an internal tool nobody wanted. The constraints turned out to be more interesting than the freedom I thought I was after. When a payout is wrong, someone loses money and somebody has to explain it. That makes teams honest.
Consumer work taught me to measure. Platform work taught me that the measurement usually points at a structural problem rather than a visual one. I use both.
Four things I do before I open a design file.
Most of the difficulty in platform work is structural rather than visual, so this is where the time goes.
Read the data model
If I cannot explain the events behind a screen, I am not ready to design it. The schema shapes the interface as much as the layout does.
Find who actually decides
An engineer and a commercial owner ask different questions. Ordering a flow by who answers each part solves more than reordering fields.
Name the tradeoff out loud
Every structural choice costs something. Saying what, before shipping, is what makes it defensible six months later.
Go back and audit it
My own work is where I find the most useful problems. One project’s whole roadmap came from auditing what I had shipped.
Opinions I have argued for, and occasionally lost.
Defaults are the real design
Most people never open advanced settings, so the default becomes the actual product behaviour.
Sixty settings, and the ones that mattered most were the ones nobody touched.
Friction is a tool, not a failure
Removing steps is the instinct. But an action that changes what real users see, and cannot be undone, deserves resistance. I argued for a dialog against a PM who wanted it linear.
One error cost more to unpick than a second of friction costs to add.
A design system is an agreement
Tokens are where the agreement gets written down, not the agreement itself. Naming by role rather than value is what made engineers reach for them.
Nobody reaches for indigo 800. They reach for primary dark.
Pull it if it does not hold up
I shipped a focus ring that looked right in isolation and broke once nested in real containers. Pausing beat leaving it half working.
Shipping accessibility theatre is worse than shipping nothing.
A short version of the path.
Senior product designer, adjoe
Platform UX across two dashboards, plus ownership of the design system. Built the user investigation platform, rebuilt publisher onboarding, and designed two new promotion types.
Senior product designer, Insider
Led the Mobile Suite team from ideation through release. Ran more than thirty usability tests and tightened design ops, which moved handoff satisfaction from 7.2 to 8.8.
Product designer, Getir
Launched car rental inside the super app and shipped EU focused features. Consumer work at speed, where the feedback loop is short and the numbers arrive quickly.
Designer, Koç University and freelance
Branding and interface work for startups and NGOs, plus a campus app redesign that cut IT support calls by more than half.
Complexity that has a cost attached
Every role I have stayed in had one thing in common: being wrong was measurable. That is the environment I look for.
Numbers I can walk you through.
Measured results from work across ad tech, martech and consumer apps. I am happy to explain how each was captured, including where the measurement is softer than the figure suggests.
Three things I spend time on.
Playing with AI
Mostly through OpenCode with Opus 5 and GPT 5.6, checking whether our design system is legible to something other than a person.
Drawing and watercolour
Art time that is only for me. No brief, no review, no system.
Family
Time with my daughter and my wife. We travel often, mostly to explore somewhere new.
If the problem is genuinely complicated, I am interested.
Open to staff and lead roles in B2B and platform teams. Happy to talk through any of the work in detail, including the parts that did not go well.
