RealifAI | Devpost

Inspiration

With rising home prices and a more than ever difficult to find "dream home", we found people often left with checkboxes or other difficult to deal with pieces of software on housing websites. In order to combat the ever increasing difficulty in finding a home, we decided to develop a software that would make the process of finding a dream home extremely easy. By just writing down the ideal qualities of your dream home, you can find and view a number of houses that fit those requirements.

What it does

This project takes the long effort of trying to find a dream home and tosses it out the window. It allows you to write down your thoughts of an ideal home and returns several possibilities that you can look at closer through a street view and a personalized message about the home.

How we built it

We built the frontend by using React and built the backend using Flask. We used the OpenAI GPT API to process the natural language and ChatGPT sends parameters to filter into a few candidate houses based on a housing dataset. The locations (based on longitude and latitude) are placed on a React Google Maps IFrame that we grabbed from the Google Maps API. When prompted, these coordinates will be used to access Street View. A unique description of the house was also generated using ChatGPT according to the wants and needs of the person.

Challenges we ran into

There were many challenges that we ran into throughout the process of creating the project. One of the main challenges we ran into was concerning the data. There was not much publicly available data to use for our project. As a result, there was an enormous amount of time spent cleaning the data and extracting features from it. Furthermore, one of the three viable datasets had to be tossed out due to the longitude and latitude not being specific enough. Another challenge that we ran into was the Google Maps API. The developer view of the map was something that took us considerable time to get rid of. Finally, React was a very new experience for all of us, so learning it was something that was very challenging as we had difficulties with getting rid of some Google Maps components as well as trying to make the page look as good as possible.

Accomplishments that we're proud of

Given that almost our entire team were first time hackers, we are proud that we could make a project on this scale and learn as much as we did. Learning react and making a responsive page was something that all of the members of the team were proud of. Furthermore, being able to work with the data constraint and its challenges taught us about how persistent we are and how devoted we are to software engineering.

What we learned

It was our first time working on an NLP project so we all got to experience using the OpenAI GPT AI and integrating it within our software. We also learned how to use the Google Maps API and how to combine it with React to create a pleasant viewing page.

What's next for RealifAI

Adding more data from the US and other countries as well as improving the search functionality for our software are the next steps we are going to take. We also intend to deploy to Google Cloud or AWS in the future so that more people can use this software.

Built With

Submitted to

VTHacks 11

Created by

Worked on FullStack, specifically the data cleaning/processing and search functionalities. Furthermore, I worked on the React functionality for the front page and map pages as well.

CSHieuV Vu
I worked on the search page with React and the communication between the frontend and backend. I also designed the NavBar that appears on every page. This was my first time using React!

Aidan Liu
I worked on FullStack development. It was my first time using the GPT API, and I learned a lot about prompt-engineering, including few-shot and zero-shot learning. I also developed some of the data processing code on the frontend, and I learned a lot more about React.

Ethan Ermovick
I worked mainly on the integration of the Google Maps API with the React front end. I worked on communication between the backend and frontend. I also worked on some backend functionality.

Ethan Coon