An app 800,000 people open every month. At the till to show their loyalty card. On the couch to buy a pyjama. HEMA's app draws its screens from the server, so HEMA can change the shop without waiting for an app release.
My part. I was the lead iOS developer, responsible for the technical vision of HEMA's iOS app. As its feature owner I also did a product owner's and an analyst's work: I owned the quarterly roadmap and weighed new features against technical improvements.
I set up HEMA's design system from scratch, and became the team's expert in server-driven UI.
A new look, an old foundation. Then HEMA refreshed its whole brand. The iOS app was built in UIKit, and the new design was a chance to move to SwiftUI. Rewriting the app was not an option. People were using it every minute of the day.
We moved it one component at a time, behind a switch, without ever freezing the app.
Where it hurt. Four of us moved 58 components over 442 days. SwiftUI inside UIKit's lists had sharp edges: state that vanished when a cell was reused, cells that would not grow when their content did, and a grid inside a list that froze the app in production. Each one taught us a rule we wrote down.
The moment it clicked. Designers started joining a weekly session around live previews in Xcode, and it grew into a shared component library. Then a product owner watched a ticket go from created to a tester's phone in sixty minutes. After that, nobody had to be convinced.
I wrote the whole journey up for the Q42 engineering blog.