Handshake's product surface spanned three distinct audiences — students, employers, and universities — each with its own visual identity, built by separate teams. The same primitives were being re-implemented in parallel, drifting apart in accessibility, behavior, and look. Design reviews were spent relitigating decisions that had already been made somewhere else.
The bet was a single component tree that could render three complete themes without a single if (brand) branch — and an adoption model with no mandate. Teams had to choose Rosetta because it was better, not because a memo said so.
Every component followed the same contract: tokens in, markup out, no brand awareness. Variants were closed enums instead of boolean sprawl. Accessibility was baked into the primitive, not left to the consumer. The library shipped with its own linting rules, codemods for every breaking release, and a documented public roadmap.
Teams had to choose it because it was better, not because a memo said so. That constraint shaped every API decision.
Rosetta became the default choice for new surfaces across the organization — 135+ components, three complete themes, adopted by teams supporting 160+ engineers. It outlived team reorgs, brand refreshes, and its author's tenure, which is the only metric a platform bet really has.
Building something at the frontier, or hiring the person who will? Reach out.