The Challenge
As the first design hire at Lyra Health, I walked into a codebase with no design consistency — every page looked like it was built by a different person (because it was). The engineering team was moving fast, and design debt was accumulating faster than features.
My Role
I was the sole designer responsible for everything visual — from product design to marketing to the design system itself. I had to build infrastructure while simultaneously shipping features.
The Approach
Start with an Audit
I catalogued every color, font size, spacing value, and component variant in the existing product. The results were sobering: 47 unique grays, 12 button styles, and zero shared components.
Principles Over Perfection
Instead of trying to build a comprehensive system upfront, I focused on three things:
- Establish foundations — A constrained color palette, type scale, and spacing system
- Build the most-used components — Buttons, inputs, cards, and modals
- Document as you go — Every component got a usage guideline
Key Design Decisions
Token-Based Architecture
I built the system on design tokens from the start, knowing we'd need to support theming and platform variations as the company grew. This investment paid dividends when we launched our mobile app.
Collaborative Process
I paired with engineers to build components together, ensuring the Figma components matched the React components 1:1. This eliminated the "design-to-dev translation" gap.
Progressive Adoption
Rather than mandating a big-bang migration, I created a migration guide that let engineers adopt the system incrementally, screen by screen.
Impact
- Reduced design-to-development time by 50% for new features
- Consolidated to 5 grays from 47, 3 button variants from 12
- 100% adoption across the product within 6 months
- System scaled to support 3 platforms (web, iOS, Android)
Reflections
Building a design system at a startup taught me that systems work is fundamentally about people, not components. The technical architecture matters, but adoption comes from empathy — understanding engineers' workflows, respecting their constraints, and making the system genuinely easier to use than the alternative.