Our website implements a secure login system for user authentication.
Homeview can take user specific queries into consideration.
Homeview even takes locations into account!

Visit Our Site, Homeview!

Test User Credentials

Username: user@homeview.studio; Password: homeviewpassword

Inspiration

We are a group of close friends who are very passionate about machine learning and its application in real problem spaces. Marketplaces like homes.com and apartments.com have helped millions of consumers to make wise financial and lifestyle decisions by finding a home that is right for them. In addition, Generative Artificial Intelligence (AI) has demonstrated itself to be a powerful technology when used to the benefit of it's users. For our hack, Homeview, we decided to explore the possibilities of how Generative AI could further the goal of bringing equitable housing to low-income communities.

PropelAuth

PropelAuth surprised us! Our team was genuinely impressed by how easy it was to set up user authentication and account management. The members of our team had never implemented user authentication in a hackathon project because, well, user authentication is so complicated that it itself could be a hackathon project! We also appreciated that it includes a reverse proxy that we could easily use to route traffic to our web application. We learned that PropelAuth has testing, staging, and production environments for each stage of development in a web application. We could have kept our login environment in the testing stage, but we wanted to have a completely professional finish on our project. We had to do some server-side and domain configuration to get the production environment to work. We were very pleased by how it turned out, and we will definitely be using PropelAuth again!

What it does

Current real estate marketplace experiences are defined by search queries parameterized by a wide set of criteria. These systems operate well, and could be made even better by integrating Generative AI to assist a variety of users. Large Language Models (LLMs) may be used as a way of increasing accessibility to the platform, saving users' time, and delivering highly personalized suggestions for each user, all through an intuitive chat interface.

How we built it

In the span of 36 hours, we brainstormed, designed, and implemented a guided real estate marketplace experience as Homeview, a full stack web application. We leverage several real estate datasets with information on house sizes, prices, and amenities, in order to cover the widest range of information a mindful user could possibly want. We are running a custom Python web server on a virtual private server (VPS) hosted in the cloud. We designed an intuitive user interface on our front end with custom styling and graphics in order to make our tool as easy to use as possible.

On the backend, we use Open AI's GPT-4o model API, which serves as the core of our user personalization system. In order to give this AI knowledge of real estate in order to benefit the user, we used a machine learning technique called Retrieval-Augmented Generation (RAG), In order to enable the model to most effectively retrieve real estate data. RAG works by first translating the raw data from dataset formats into vector embeddings, which are a compressed and symbolic form of data storage that the model can understand. We store these vector embeddings in FAISS, Facebook's vector database designed for machine learning applications. Our model can then readily retrieve any particular real estate data with efficiency to provide information to the user. We may sound like a broken record here, but this technology is incredible! It is so exciting to be able to apply higher-level mathematics to problems in the real world.

Challenges we ran into

We wanted to implement Mixtral 7B, which is notable for outperforming Meta's Llama 1 and Llama 2 on many benchmarks. Due to computing limitations, we weren't able to query responses fast enough, and instead pivot to using Open AI's GTP-4o model API. We also initially struggled to find the right balance between maximizing the data available to our model and the amount of time we had. There are plenty of datasets describing quantifiable aspects of real estate, but we weren't really satisfied with just getting the numbers. A house might have a great price and the perfect number of bedrooms, but there could be other critical information on, say, a homes.com posting, that a numerical dataset wouldn't account for. We ultimately settled on using an open source solution to aggregate results from many different real estate marketplaces to create a more accurate consensus. We worried that we did not have time to implement such a complex mechanism, but we were able to finish it just in time before the hackathon ended.

Accomplishments that we're proud of

Implemented a complex Machine Learning system, Large Language Model, Retrieval-Augmented Generation, Live Document Embedding, and Vector Database within the span of a weekend.
Created a beautiful website, with the help of PropelAuth on the user authentication interface, and otherwise with a lot of time spent on making the design as intuitive as we could imagine.
Developed teamwork through shared development environments. Our team used developer tools such as tmux to collaboratively diagnose and debug technical issues while looking at the same screen from different computers. We started doing this accidentally, but it ended up being a creative way to work together that we had a lot of fun with!

Skills we developed

Artificial Intelligence / Machine Learning Web backends in Python & Node.js User Interfaces / User Experience Server Administration, Writing secure firewall rules, Writing DNS Records

What's next for Homeview

We are so proud of what we accomplished in 36 hours, and we also hope to continue working on Homeview to see the impact that Generative AI could have on the real estate industry. We see many potential advancements helping consumers make wise financial decisions and find homes that they fall in love with. Some ideas we want to continue to explore are:

Implement even more powerful AI models.
Trials for people actively looking for homes.
Build a relationship with the CoStar Group to potentially build a partnership and make AI technology accessible to consumers.

Thank you so so much!

We had a great time this weekend! We really enjoyed meeting other hackers, talking to the event sponsors, and attending workshops. We look forward to signing up again next year!