README.md

---
title: Property AI
emoji: 🏠
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: "3.44.0"
app_file: app.py
pinned: false
---

# NoBrokerage Chatbot

**AI-powered real estate assistant** for NoBrokerage.com that retrieves property data from a FAISS vectorstore, applies structured filters, and generates grounded summaries and property cards using Groq LLM.  

This project is built using **FastAPI**, **LangChain**, **FAISS**, **HuggingFace embeddings**, **Frontend**,  and **Groq LLM**, and is ready for **Docker deployment** and **Hugging Face Spaces**.

---

## Table of Contents

- [NoBrokerage Chatbot](#nobrokerage-chatbot)
  - [Table of Contents](#table-of-contents)
  - [Project Overview](#project-overview)
  - [Features](#features)
  - [Project Structure](#project-structure)
  - [Example Queries the Chatbot Can Handle](#example-queries-the-chatbot-can-handle)
  - [How to Run Locally](#how-to-run-locally)
    - [1. Clone the Repository](#1-clone-the-repository)
    - [2. Create Virtual Environment](#2-create-virtual-environment)
    - [3. Install Dependencies](#3-install-dependencies)
    - [4. Set Up Environment Variables](#4-set-up-environment-variables)
    - [5. Run the FastAPI Server](#5-run-the-fastapi-server)
  - [Tech Stack](#tech-stack)
  - [Features](#features-1)
  - [Deployment Ready](#deployment-ready)
  - [👨‍💻 Author](#-author)

---

## Project Overview

NoBrokerage Chatbot allows users to query property listings by specifying filters like **city, BHK, budget, status, locality**, and returns **summary text** and **cards** with property details.  

- **Semantic search**: FAISS vectorstore with HuggingFace embeddings for similarity search.  
- **Deterministic filters**: Apply structured metadata filters for city, BHK, budget, status, and locality.  
- **LLM summarization**: Groq LLM produces grounded summaries and card outputs strictly from filtered property records.  
- **Deployment-ready**: Can run via CLI, FastAPI, Docker, or Hugging Face Spaces.

---

## Features

- Parse natural language queries for:
  - Budget (₹, Cr, Lakh)
  - BHK
  - City
  - Property status (Ready to move / Under construction)
  - Locality or project
- FAISS similarity search over property embeddings
- Deterministic filtering of search results
- Generate structured JSON output with:
  - `summary` (text summary)
  - `cards` (detailed property info)
- FastAPI backend with `/chat` endpoint
- Dockerized for easy deployment
- Compatible with Hugging Face Spaces

---

## Project Structure


```bash
NOBROKERAGE/
├── backend/
│   ├── api.py
│   
├── data/
├── database/
├── frontend/
├── processed_data/
├── src/
│   └── chatbot.py
├── subha/
├── vectorstore/
│   └── index.faiss
├── .env
├── .gitignore
├── Dockerfile
├── README.md
└── requirements.txt

```
---

##  Example Queries the Chatbot Can Handle

The chatbot can intelligently respond to natural language queries like:

- " Find 2BHK apartments in Chembur "
- " 3BHK flat in Pune under ₹1.2 Cr "
- " Under-construction 3BHK in Mumbai "
- " 2bhk flat in pune "
- " 3bhk in Mumbai "
  

It uses:
- **FAISS** to find the most relevant property documents.
- **LangChain + Groq LLM** (`llama-3.1-8b-instant`) to summarize matching results.
- **Structured filters** for city, budget, BHK, locality, and status.
---

##  How to Run Locally

### 1. Clone the Repository
```bash
git clone https://github.com/yourusername/nobrokerage.git
cd nobrokerage
```
### 2. Create Virtual Environment
```bash
python -m venv venv
venv\Scripts\activate   # on Windows
# OR
source venv/bin/activate  # on Mac/Linux
```
### 3. Install Dependencies
```bash
pip install -r requirements.txt
```
### 4. Set Up Environment Variables
```bash
GROQ_API_KEY=your_groq_api_key_here
```
### 5. Run the FastAPI Server
```bash
cd backend
uvicorn api:app --reload
```
---

##  Tech Stack

- **FastAPI** — Backend API framework  
- **Frontend** - index.html, style.css, script.js
- **LangChain** — LLM orchestration  
- **Groq LLM (llama-3.1-8b-instant)** — Summarization & reasoning  
- **FAISS** — Semantic vector search  
- **HuggingFace Sentence Transformer** — Embeddings  
- **Docker** — Containerization  
- **Python 3.11**

---

##  Features

✅ Semantic property search using FAISS  
✅ Intelligent summaries and cards generated by Groq LLM  
✅ Handles filters like city, budget, BHK, and project status  
✅ Ready for Hugging Face Spaces or cloud deployment  
✅ Modular architecture (backend + src separation)

---

##  Deployment Ready

This backend is designed to work seamlessly with **Docker** and can deploy directly to **Hugging Face Spaces**.  
Make sure `vectorstore/` and `.env` are included in your project before building the Docker image.

---

## 👨‍💻 Author

**Subhakanta Rath**  
🎓 MSc AI & ML — IIIT Lucknow  
💡 Focused on ML, Data Engineering & Agentic AI Systems  
📍 Lucknow, India