I have built three design systems from scratch. Nukta at SadaPay. Xperia during an earlier role. Techtonic at ZAR. Each one taught me something different, but the lesson that ties them all together is this: a design system that cannot survive without its creator is not a system. It is a personal preference document with component names.
Nukta was the first system I built at scale. At its peak, it powered an app used by over five million people and was maintained by a team of eight designers and a dozen engineers. When SadaPay was acquired by Papara, I left — and Nukta kept running. That was the point. The tokens were documented. The components had clear usage guidelines. The naming conventions were predictable enough that a new designer could pick up a file and understand the logic without asking anyone. I did not build Nukta to be impressive. I built it to be obvious.
I did not build Nukta to be impressive. I built it to be obvious.
Techtonic at ZAR was a different challenge. We were building for 110 countries, multiple currencies, and a user base that ranged from crypto-native traders to first-time digital wallet users. The system had to be flexible without being chaotic. Every component needed to work in English, Arabic, Spanish, and Swahili. Every spacing decision had to account for text that might be 40% longer in German. I learned that the hardest part of a multi-market design system is not the technical architecture — it is the governance. Who decides when a component gets a new variant? Who says no? Without clear ownership and decision-making frameworks, a design system becomes a junk drawer of edge cases.
The throughline across all three systems is the same: the goal is not to build something beautiful. The goal is to build something that makes the next person's job easier. A design system should reduce decisions, not add them. It should feel like a constraint that frees you up, not a rulebook that slows you down. The best test of a design system is not how it looks in a Figma library. It is what happens six months after the person who built it walks out the door. If it is still being used — still being extended, still making sense to people who never met you — then you built something real.