Belong-Roomate_Finder

Prachi Prajapati — Mon, 05 Oct 2026 09:59:37 +0000

What I Built

I built Belong, a roommate and flat-finding platform for my best friend Shalini, who was struggling to find a place to live with a roommate who was actually compatible with her lifestyle.

Most rental platforms focus mainly on location, rent, and room type. Belong goes a step further by considering lifestyle preferences such as sleep schedule, smoking, pets, guests, cleanliness, and social preferences.

Users can enter their preferences, discover suitable listings, see a compatibility percentage, and ask Gemma, the AI assistant, to explain why a particular place or roommate is a good match.

The goal is simple: help people find not just a place to live, but a place where they truly belong.

Demo

Frontend: https://belong-roommate-finder-frontend.onrender.com/
Backend API: https://belong-roommate-finder.onrender.com

Code

GitHub: https://github.com/prachi-p-jpg/belong-roommate-finder

How I Built It

I built Belong as a full-stack web application using React + Vite for the frontend and Node.js + Express for the backend. User data, preferences, listings, and saved homes are stored using MongoDB Atlas.

The core matching system calculates a compatibility percentage based on the user's preferences, including budget, location, room type, sleep schedule, smoking, pets, guests, cleanliness, and social preferences.

I also integrated an AI assistant called Gemma to help users understand their matches. Instead of simply returning a generic chatbot response, the AI receives the calculated compatibility information and explains why a listing is a good or bad match, including the similarities and differences between the user's lifestyle and the listing.

The application is deployed with the React frontend and Node.js backend running separately, with the AI API key kept securely on the backend.

Why Does Open Innovation Matter?

Open innovation made it possible for me to build an AI-powered feature as a beginner developer without having to train an AI model from scratch.

Instead of using AI only as a general chatbot, I connected it to Belong's own matching logic. The application first calculates a compatibility score from structured user and listing preferences, and then the AI explains that result in simple, human-friendly language.

This approach allowed me to combine traditional application logic with AI to solve a real problem: helping someone understand not only which place matches them, but why it matches them.

It also made the project more accessible to me as an independent developer, because I could experiment with AI-powered functionality without needing the resources required to build and train a model from the ground up.

My Agent Session

I used an AI-assisted development workflow to build Belong, including planning, frontend development, backend development, API integration, debugging, and deployment.

Prize Categories

Best Use of Render — Belong's React frontend and Node.js/Express backend are deployed on Render, making the full-stack application publicly accessible.
Best Use of Gemma — Belong uses an AI assistant to explain roommate and listing compatibility based on the user's lifestyle preferences and calculated match score.
Best Use of MongoDB Atlas — MongoDB Atlas is used as the application's data layer for storing users, preferences, listings, saved homes, and other application data.

Final Thoughts

Building Belong was a great learning experience for me. I built it with the goal of helping my best friend Shalini find a place where she could feel comfortable and truly belong.

This challenge helped me learn more about AI, full-stack development, deployment, and building a project around a real-world problem.

Thanks for checking out Belong! ❤️

DEV Community: Prachi Prajapati