Case study · SEDA · 2026
I built SEDA's first design system in 2024 to give the explorer one language. In April 2026, I rebuilt it: calmer, sharper, and ready for products customers would trust with their keys.

The 2024 system did its job: it gave the explorer a consistent look and let the team ship new pages fast. But it spoke crypto: neon purple, all-caps labels, glowing gradients.
SEDA Fast was a different kind of product: a dashboard where customers manage the keys to their data. It needed to feel professional and trustworthy. It was also self-serve, so it needed components the explorer never had: richer charts, toasts, and the building blocks for billing and team management.


Two tracks running in parallel, so engineers never waited on a finished system.



Three decisions, and each one made the next easier.
If docs are useful, people use them. If they're confusing, no one does. I documented the whole system in Notion, from foundations down to every token and value, so designers and engineers could move fast without asking.
Pages link to each other, so a spec never stands alone: a component that uses core sizes links straight to the core sizes page. Every page carries visual references and links back to its source in Figma, so nobody has to guess what a component should look like.
When a new engineer joined, he told me the docs gave him everything he needed to build the system.
And one system now carries both SEDA's product and its public face.
The revamp came out far better than the original, and the reason wasn't skill. In 2024 there was no time to stop and ask why. I had to start shipping components straight away. In 2026 I had room to think, and to pitch a direction before building it.
That's the lesson I keep: give designers time to understand why a choice is right, and the system holds together on its own.