Open-source healthcare UX

Balancer: UX/UI design for an AI healthcare tool

Balancer is an open-source clinical decision-support tool created by volunteers through Code for Philly. The product helps clinicians research evidence-based medication options for bipolar disorder.

I contributed UX/UI design across information architecture, accessibility, workflow clarity, and design-system thinking, while also supporting a special project focused on chatbot guidance and response safety.

  • Role: UX/UI Designer
  • Organization: Code for Philly
  • Project type: Volunteer, open-source healthcare tool
  • Focus areas: Healthcare UX, accessibility, information architecture, AI guidance

Led UX workshops and design reviews with a cross‑functional volunteer team

  • Improved clarity, trust, and navigation for a healthcare experience with sensitive content
  • Contributed sitemaps, accessibility improvements, style‑guide foundations, and front‑end collaboration support
  • Helped shape safer, more consistent chatbot behavior for clinician use

The Challenge

Designing for trust in a high‑stakes context

Balancer was not a typical interface problem. The product had to support clinicians working with complex medical information, which meant the experience needed to feel clear, credible, and easy to navigate. Because the project was volunteer-built and open source, the work also had to support lightweight handoff, practical collaboration, and future scalability. At the same time, the team was exploring AI‑assisted interactions. That raised an additional UX challenge: the interface and language needed to be useful without sounding overconfident, and structured without creating unnecessary friction.

  • Clinicians needed fast access to readable, trustworthy information
  • Medical content required a careful tone and clear communication
  • The experience combined traditional UX issues with AI behavior design
  • The team needed lightweight processes that fit an open-source workflow
  • Design decisions had to support consistency as the product evolved

Trust sensitive content · Accessibility needs · Volunteer collaboration · Scalable handoff

What I changed

UX Improvements and Team Contribution

My work focused on making the product easier to understand, easier to navigate, and easier for the team to build consistently over time. I contributed both to the user experience itself and to the structure around it, including workshops, documentation, and collaboration with front‑end contributors.

  • Led UX design workshops and cross‑functional reviews to align the team on priorities
  • Conducted a content audit to identify trust gaps and improve the clarity of medical information
  • Created public‑facing sitemaps to simplify user flows and reduce navigation friction for clinicians
  • Initiated a Balancer style guide to support UI consistency and future scalability
  • Logged and organized front‑end requests in GitHub Kanban to improve collaboration between design and development
  • Supported research planning by helping identify patient participants for interviews

Special Project:

AI Chatbot UX and Safety

In addition to product UX work, I led a special project focused on chatbot behavior and response safety. The goal was to turn abstract product requirements and safety concerns into practical rules the team could use. I helped define how the chatbot should sound, when it should escalate, and how starter prompts could help clinicians begin using the tool more confidently. This work was especially important because the chatbot needed to feel helpful without overreaching.

The solution was not just a better interface.

It was a better governance structure for tone, prompts, and review.

My role was to make the chatbot more usable and more governable at the same time.

  • Defined the core rule set for chatbot behavior, tone, and response safety
  • Turned product requirements and safety constraints into practical decision rules
  • Added escalation guidance for cases that required human professional support
  • Helped launch approved starter prompts so clinicians could begin quickly without a larger admin workflow
  • Standardized response language to improve consistency, respectfulness, and risk control
  • Created a lightweight expert-review process so prompt updates could be governed, tracked, and improved over time