●Built an AI-powered tool that helps patients understand and challenge medical bills by uploading bills and extracting charges within seconds.
●Implemented a Fairness Score system that summarizes how clean the bill is, providing transparent and explainable scoring that reflects 'worth reviewing' rather than correctness.
●Created visual highlighting system (red and yellow) that identifies potential issues like duplicate charges, vague descriptions, temporal inconsistencies, and place-of-service inconsistencies.
●Developed an interactive Q&A feature where users can ask ClearBill why something was flagged, receiving clear explanations in plain language with full context.
●Implemented one-click dispute letter generation that creates professional, neutral-toned letters requesting clarification or itemization, including user guidance on who typically receives the letter.
●Architected the full-stack application using Next.js 14, TypeScript, Tailwind CSS, Supabase (Postgres + Storage), and Gemini (text + vision) for OCR and analysis, deployed on Vercel.
