CareCompass

About the Project

CareCompass is an AI-powered web app designed to help people find nearby health and support resources based on their needs. The idea came from a simple but important problem: when people are under stress, confused about what kind of help they need, or facing a difficult health-related situation, it can be hard to know where to start. Searching online often gives scattered or overwhelming results, especially for people who need quick and practical support. We wanted to create a tool that could make that process easier by turning natural language into useful local guidance.

What inspired us most was the possibility of building something that could have a direct community impact. Instead of making a project that was only technically interesting, we wanted to create something meaningful and realistic. We focused on the Merced area first so that the project would stay manageable and grounded in a real local setting. This helped us think about the app not just as a demo, but as a tool that could eventually be expanded to help more communities.

Through this project, we learned a lot about both technical development and project execution. On the backend side, this was my first time working with APIs in a real project. I learned how to structure a FastAPI backend, create routes, handle requests and responses, and connect an AI model to the application so user input could be analyzed in a more flexible way. I also learned how to move from a basic idea into an actual working product by breaking the problem into smaller parts: frontend, backend, data, AI classification, and map visualization. My teammate focused on the frontend and styling, which helped transform the project from a functional prototype into a more polished and user-friendly experience.

We built the project using React + Tailwind CSS for the frontend, FastAPI for the backend, Leaflet for the map interface, and the OpenAI API for natural language understanding. The workflow is simple: a user types in what kind of help they need, the backend sends that text to the AI model, the model classifies the need into categories such as mental health, primary care, food assistance, or emergency care, and then the app returns relevant local resources on a map and in a results list. We started with a curated dataset for Merced so that the system would stay reliable and easy to demo.

One of the biggest challenges we faced was turning a broad idea into something realistic for a hackathon. At first, the concept felt much bigger than what we could build in a short amount of time. We had to narrow the scope, limit the project to one area, and focus on a minimum viable product. Another challenge was figuring out how to integrate AI in a way that was actually useful instead of just adding it for the sake of saying the project used AI. We decided that the best role for AI was understanding natural-language input and translating it into structured categories, rather than generating the resource data itself.

Another challenge was technical setup. Since this was our first time integrating APIs and AI into a program, We had to learn how to configure the backend, connect the frontend to it, pass data correctly between files, and debug issues with Tailwind, React, and request handling. OpenAI helped us a lot throughout that process by helping me break down the project structure, understand how to connect the backend and frontend, and work through the smaller implementation steps needed to turn the original idea into a functioning application. In that way, OpenAI was not just part of the final product, but also part of the learning and building process that helped make the project possible.

Overall, this project taught us how to balance ambition with practicality. We learned how to collaborate across frontend and backend roles, how to use AI as a meaningful feature, and how to turn an idea focused on community support into a working prototype. CareCompass represents both a technical learning experience and a project we believe could grow into something with real-world value.

Built With

• fastapi
• openai
• react-leaflet
• tailwindcss