Design System for Global MedTech

Core components had to support very different clinical workflows and continue evolving as product requirements changed. Designers needed the system to help them move, not force them to invent around it.

I designed and evolved components based on feedback from product teams and consumers of the system. That included complex patterns such as timeline steppers, where state, sequence, hierarchy, and progressive disclosure all had to work together. The goal was not to satisfy one screen. It was to turn a specific need into a reusable pattern without weakening the rest of the library.

Healthcare interfaces are inherently information-dense. Clinicians often work for long periods under pressure, moving between records, images, actions, and decisions. Too much space can hide relationships and force unnecessary navigation; too little can make every element compete for attention.

I piloted a density and spacing strategy across layout, spacing tokens, and component behavior. The work explored how the system could support more compact modes while protecting readability and focus. It laid the groundwork for flexible layouts that adapt to the task instead of applying one comfortable-but-generic density everywhere.

Illustrations and icons were being created with external agencies. Without close art direction, small differences in metaphor, detail, and tone could accumulate into a fragmented visual language. This was particularly risky when imagery also needed to communicate clearly in a medical context.

I worked with agency partners early and iteratively, reviewing style, clarity, and fit with the wider system. That feedback loop produced assets that were not only visually coherent, but reusable and maintainable across products rather than tied to a single campaign or surface.

I led recurring design reviews focused on component integrity, system alignment, and overall UX quality. A review was never limited to checking whether the correct component had been used. I applied usability heuristics and accessibility principles to identify problems that strict component compliance alone would miss.

This changed the conversation from policing a library to improving the product. Teams left with specific guidance, the Design System gained evidence about recurring gaps, and both sides became better at recognizing when a local problem should lead to a system-level improvement. I also initiated and co-developed Figma plugins and widgets to remove repetitive friction from these workflows.

The Design System backlog kept growing while stakeholder pressure and priorities shifted. With limited capacity, the team needed a way to demonstrate progress without lowering the quality bar or letting the loudest request become the next priority.

I created a “Quick Wins” filter and introduced impact-versus-effort triage to make prioritization easier to understand and defend. I also strengthened the Definition of Done and improved user stories and acceptance criteria. These changes reduced ambiguity before work entered delivery and made trade-offs clearer to both the team and its stakeholders.

Design decisions affected regulated medical software, where errors can carry clinical and legal consequences. Accessibility and usability could not be treated as a final audit added after the interaction model had already been decided.

I partnered closely with the regulatory usability team to validate components and interaction patterns against medical usability standards and WCAG AA accessibility requirements. Bringing that expertise into the design process helped us resolve risk earlier and protect both design quality and reliability in high-stakes clinical environments.

Small inconsistencies scale quickly in a Design System. A missing edge case or unclear interaction can be implemented several different ways, then copied into every product that adopts the component.

I worked directly with the in-house engineering team responsible for building the system in Flutter. During structured refinement sessions, I clarified behavior, states, edge cases, and micro-interactions before implementation began. I also joined product and cross-team discussions to provide guidance grounded in Flutter's constraints and capabilities. This made the translation from design to production more predictable and prevented ambiguity from becoming platform-wide inconsistency.

GenAI features were shipping faster than the platform could establish a shared look and behavior for them. Without direction, they risked feeling bolted on. Motion decisions were ad hoc, and the experience designed in Figma could drift from the one implemented in code. This was an especially costly loss of trust in a clinical product.

I took creative direction for the GenAI surfaces and translated the visual language into tokenized, buildable components. I defined the timing and easing behind key moments so motion communicated state and intent rather than acting as decoration. I then prototyped Flutter components in an AI-assisted workflow and paired with engineering, bringing the designed and shipped experiences closer together from the start.

The team did not need another AI experiment to try once and forget. It needed tools that performed real work and could be trusted around a system of this scale.

I built reusable AI skills for recurring tasks. Jira ticket creation went from minutes to seconds. A Confluence skill reduced the effort of producing documentation. I also built a Figma plugin that finds detached components and ghost variables across files containing 10,000+ layers. Instead of adding AI beside the process, I reworked the process so AI had a clear, useful place inside it. Today, a cross-functional team of roughly ten designers, developers, and product colleagues uses these tools in its everyday work.