Tokenized Design System

Overview

This case study traces the evolution of my design systems practice—from early asset abstraction to fully tokenized systems—showing how foundational system thinking enabled scalability, theming, and behavioral consistency across products, platforms, and organizations.

Role

Design systems lead and contributor across multiple organizations. Defined system abstractions, contracts, and governance. Partnered closely with engineering to align design intent with implementation. Mentored designers and guided system adoption and evolution.

System / Work

The work began at Roku, where supporting dozens of OEM partners required the system to handle frequent icon swaps and theming changes driven by manufacturer deals. XML-based abstraction was introduced to separate assets from implementation, enabling replacement without rewriting the system, though tooling limitations made theming and library management difficult to maintain.

At Google, the emergence of Sketch enabled a tighter bridge between design and engineering. Vector-based assets defined in JSON allowed shared SVG libraries and live specs, reducing translation loss and enabling a common source of truth.

At Amazon, the work matured into a robust tokenized design system, particularly for automotive platforms. Tokens were structured around base and semantic layers, allowing typography, color, layout, and theming to adapt through rules rather than manual updates. This approach enabled system-level capabilities such as restriction management based on rules and conditions.

This work continues through mentorship and leadership at 42, supporting designers and teams in applying tokenized systems thinking to new platforms and use cases.

Outcome

Enabled scalable theming and OEM customization without fragmenting systems. Reduced design–engineering friction through shared abstractions. Established token-driven rules supporting adaptive layouts and behavior. Laid the foundation for higher-level systems such as global UI surfaces and embedded AI.

With a shared design language in place, the next challenge became orchestration—how multiple surfaces and applications coexist without conflict.