The White Coat Way

✨ About the Project

When we were brainstorming ideas for something impactful, I turned to the people I live and work with — my residents. As a Community Assistant at Palo Verde West, I support a lot of pre-med students, and almost every single one said the same thing: "Finding pre-med opportunities is so hard, and a lot of them are hush-hush."

That’s when the idea for White Coat Way was born — a platform where anyone, whether a high schooler or a college student, can discover research programs, internships, shadowing, and more — no connections or insider knowledge needed.

💡 What Inspired Us Hearing how many students felt lost in their pre-med journey pushed us to create something that levels the playing field. Everyone dreaming of that white coat should have access to the resources that help get them there — not just the well-connected.

🧠 What We Learned How to build a working resume parser + recommendation engine using real-world tools like Affinda and Hugging Face models How to structure backend logic to handle user inputs, parse resumes, and dynamically return personalized suggestions That building community-driven tools is just as much about the people as the tech

🛠️ How We Built It Frontend: Built with Vercel-hosted Webflow pages for a clean and aesthetic UI Backend: Flask API that parses resumes using Affinda, takes in user goals via prompt, and feeds both into the mistralai/Mistral-7B-Instruct-v0.1 model to generate personalized opportunity matches Scraper: Pulled opportunity descriptions from a static list of 15 pre-approved URLs Display: Results shown with summaries, direct links, and the model's explanation of why each opportunity was a good fit Community Forum: Built a discussion board using Vercel-hosted custom pages, allowing peer support and resource sharing

⚠️ Challenges We Faced Time Constraints: With limited time during the hackathon, we couldn't build a fully dynamic pipeline. Instead, we scraped from a static list. Integrating AI Models: Mistral gave great results, but tuning inputs/output formatting for clarity and usefulness was tricky. Resume Parsing Accuracy: Mapping unstructured resume data into usable formats sometimes required a lot of cleanup.