Inspiration
Inspired by the challenges faced by young individuals in the real estate market, we created HomeHeart to streamline the home search process. Our mission is to provide a user-friendly platform tailored to recent graduates and first-time buyers.
What it does
Introducing HomeHeart, the revolutionary real estate software designed to simplify home buying for college grads and first-time buyers. With our seamless integration of listings, user prompts, and Pinecone's powerful similarity search, personalized recommendations are just a click away.
At HomeHeart, users effortlessly input their preferences—location, budget, and features. Pinecone's indexing and search prowess then uncovers the top three listings that closely match their criteria, streamlining the search journey.
But what truly sets us apart is our inventive use of Pinecone. By blending user prompts and embedding generation, we deliver tailor-made recommendations, honoring individuality.
We obsess over user experience too. Our intuitive Vercel-powered interface ensures a seamless journey, while Clerk ensures secure authentication. Through Pinecone's similarity search, HomeHeart guarantees precise results, aligning with users' needs.
Moreover, we effectively integrate partner tools—Pinecone, AWS, Clerk, Cohere, LangChain, and OpenAI—enhancing functionality and value.
How we built it
HomeHeart combines cutting-edge technologies. The React frontend is deployed on Vercel, while Clerk ensures secure authentication. Python powers the backend hosted on AWS. We used Cohere to generate embeddings stored in Pinecone, enabling accurate similarity searches and recommendations.
Challenges we ran into
Our entire team consists of amateur developers, and most of our original team members—a set of strangers who met on the Internet—quietly dropped out without contributing. As a result, we only had time to complete and deploy the backend. On the frontend, we ran into some trouble serving props from our database that prevented us from getting across the finish line.
Accomplishments that we're proud of
Our data scraping, cleaning, and preprocessing was extensive. Among other things, we used langchain + OpenAI to help categorize fields in the rather large original dataset and to rate the likelihood that fields would be useful to the user. This enabled us to quickly narrow down the data we wanted to keep and discard.
What we learned
Our technical skills grew by leaps and bounds. We delved into the intricacies of frontend and backend development, mastering the art of crafting intuitive interfaces and building robust systems. None of us had ever worked with most of these tools before. Vector databases and semantic search were completely new to us, and langchain posed particular challenges to learn because of the state of the documentation.
In the end, our journey with HomeHeart taught us that disruptive innovation is not just about cutting-edge technology—it's about the human element. It's about fostering a culture of collaboration, open communication, and shared purpose. By embracing these principles, we were able to build a solution that simplifies the home-buying process and empowers young individuals to find their perfect homes.
What's next for Home Heart
Our future plans include enriching HomeHeart with HuggingFace's image description model for comprehensive listing information. OpenAI integration will facilitate communication with sellers. Regular updates and maintenance ensure HomeHeart remains valuable for graduates and first-time buyers. We want to continue exploring the possibilities with all the new tools being created in this space.
Log in or sign up for Devpost to join the conversation.