Symphony: Design System & Research Platform

When I joined Symphony, the MVP was already live, being used daily by their internal team for drug diffusion studies. Anexinet had delivered it with a pink/green/black identity. Then Symphony rebranded: new logo, all green. They sent us the new color palette and reference screenshots from another app they used internally, and that was the brief.

This was an internal platform, not a consumer product. Symphony's researchers use it to design and manage drug diffusion studies: defining target patient profiles, setting campaign channels (TV, internet, outdoor/rural), determining frequencies, and then tracking results through a reporting phase. Complex, domain-specific workflows with many conditional fields and interrelated data points.

The challenge: take a working platform, fully redesign its visual identity and component library from a color palette alone, without breaking the workflows medical researchers depended on. No existing design system, no UI kit. Everything had to be built from scratch.

I worked directly with the CEO and Product Owner at Symphony, with a BA serving as the main bridge on the Anexinet side, plus a development team of 7-10 people (developers and QA). Daily meetings at the start, then bi-weekly deliverables once the rhythm was established.

The delivery model was: design by week, development handoff by section. I would complete a section of screens, hand it off to dev, and move to the next. The platform stayed live throughout, so there was no pause window.

Building the design system was the first real task. Symphony gave colors and nothing else. Every button, form state, navigation pattern, table, modal, and data visualization component was designed from zero. The existing MVP screens served as functional reference, but visually almost everything changed: not just colors, also spacing, sizes, shapes, and hierarchy.

The platform had two main phases that shaped the interface structure: study creation and reporting. Study creation involved complex forms: patient demographic conditions, media channel configuration (TV, internet, rural), frequency settings, and campaign parameters. Each field had domain logic that determined what came next.

The reporting phase had its own layer of complexity: researchers would select a study, filter results by company, reach, or metric type, and generate deliverables. Within those reports, they could add additional conditions: filtering within filters.

Building components that handled all of this while feeling clean enough for daily use was the core design challenge. I coordinated weekly with the dev team to ensure feasibility before finalizing any component, since some of the conditional logic had real implementation implications.

The new visual identity and the component library were handed off section by section, and the platform stayed in use the whole time.

Deliverables: complete design system (components, patterns, states), redesigned study creation flows, reporting interface, and a clickable prototype for stakeholder review.

The main thing this project taught me was how to work in a domain I did not fully understand. Medical research terminology, study logic, campaign planning. I was never going to be an expert. The job was to make those workflows feel structured and navigable for people who were experts. That separation (domain knowledge stays with the client, design clarity is mine) was something I had to earn through the project, not assume from the start.