Designing With R&D So Accessibility Is Architecture, Not Cleanup

The fastest way to make a CTO distrust design is to throw pixel-perfect mockups over the wall and call accessibility 'a frontend concern.' In public-sector work it isn't — WCAG is a legal constraint that touches component architecture, semantics, and state management. So I bring R&D in while the structure is still soft. I design around a shared component model rather than one-off screens, because a Design System is where accessibility actually lives: focus order, semantic roles, contrast, error handling, and i18n get solved once at the component level instead of relitigated on every page. That's a conversation engineers want to be in early, and it's where their constraints make the design better — I'd rather know what's expensive to build before I commit to it. I also design the Layered Complexity model explicitly with R&D: the simple core path and the optional depth often have very different technical risk, and surfacing that lets engineering sequence the build sensibly. The mature, non-startupy feel a government service needs comes from this discipline. Trust on the outside is the product of clean contracts on the inside between design and R&D.