Inspiration

My family’s struggles with rapidly rising hospital bills and opaque billing inspired BillBuddy; my experience and values drove the need for clear, patient-facing explanations and cost estimates.

What it does

Enriches billing records with a curated glossary, translates medical jargon into plain English, and estimates likely out‑of‑pocket costs. Adds a browsable glossary card on Home for search, filter, preview, and detail views

How I built it

Modular prompts: break tasks into small, distinct prompts (extract, match, enrich, explain) so that each can be tested and replaced independently. Use combinations of local and remote servers to store healthcare information

Challenges I ran into

A poor prompt earlier produced a program within a program, so I had to restart, seeding the project with a better and defined prompt.

Accomplishments that I'm proud of

Delivered a 50‑entry plain‑language glossary and integrated it into the pipeline. Reduced manual review with a reliable merge and fuzzy fallback. Launched the Home glossary card and improved UI in BillBuddy V2 with light/dark mode support and better accessibility.

What I learned

I learned how coding works and how different programs (GitHub, Node.js, Visual Studio Code) all work in unison. I also learned how prompt engineering is not an exact science, and you must be able to excel in communication with your language model.

What's next for BillBuddy

Expanding the glossary by at least 100 terms, add user-specific information (ex. medical records, prescriptions, emergency contacts), add real data from actual practices, making BillBuddy the place for all of your health needs.,

Share this project:

Updates