Personalized Real Estate Agent

HomeMatch is a real estate matching system that leverages natural language processing (NLP) and vector databases to provide personalized home listings. The project utilizes OpenAI’s models alongside LangChain to process user preferences and match them with synthetic real estate listings. The system generates tailored home descriptions, ensuring that each recommendation aligns with the buyer’s specific needs.

The key concepts are shown in project.ipynd and a slightly more refined CLI application can be run using the HomeMatch.py file (see below for further instruction).

Local Dev Installation Guide

This project uses Conda for environment management and Makefile commands for easy setup. Follow the steps below to get started.

1. Install Conda (if not already installed)

If you don’t have Conda installed, download and install Miniconda or Anaconda.

Check if Conda is installed:

conda --version

2. Create the Conda Environment

Run the following command to create a new Conda environment with Python:

make create_env

This will create a Conda environment named my_env (or another name if modified in the Makefile) with Python installed.

3. Activate the Environment

Once created, activate the Conda environment:

conda activate my_env

4. Install Rust (If not already installed)

Some packages, like tiktoken, require Rust for installation. Run:

make install_rust

This will check if Rust is installed and install it if necessary.

5. Install Required Python Packages

Now, install all dependencies listed in requirements.txt:

make install_reqs

This will:

• Ensure pip, setuptools, and wheel are up-to-date.
• Install all dependencies from requirements.txt.

6. Add env to Jupyter Notebook Kernal

In order to use the enviroment we created with the make commands, while the env is active run

python -m ipykernel install --user --name=my_env --display-name "Python my_env" 

then when you start up the notebook choose the Python my_env for the Kernal

Running CLI Application

With the proper environment activated with conda run

python HomeMatch.py

This will start the program and will prompt you with serveral questions such as:

1. The path of your vector database.
2. If you would like to generate new synthetic listings

From there you will be prompted with questions about what home you're in the market to buy and the application will then query over the synthetic data and return a personalized description of the property to you.

Example Outputs

Example 1

Example 2

Example 3