Apartments - Real Estate Search Platform

A full-stack real estate search platform with AI-powered chat interface and semantic search capabilities. Built with React, TypeScript, Node.js, and MongoDB.

🚀 Features

For Users

• AI-Powered Chat Interface: Natural language search for apartments using Hebrew language support
• Semantic Search: Vector-based search using embeddings for intelligent property matching
• Advanced Filtering: Filter apartments by city, price range, rooms, size, and more
• Property Details: Detailed view of each apartment with images and specifications
• Responsive Design: Modern UI built with Tailwind CSS and Radix UI components

For Administrators

• Admin Dashboard: Protected route for managing apartments
• CRUD Operations: Create, read, update, and delete apartment listings
• Authentication: Secure login system with JWT tokens

🛠️ Tech Stack

Backend

• Node.js with Express.js
• TypeScript for type safety
• MongoDB with Mongoose ODM
• JWT for authentication
• bcryptjs for password hashing
• @xenova/transformers for embeddings (multilingual-e5-base model)
• @langchain/ollama for AI chat (gemma2 model)
• express-session for session management

Frontend

• React 18 with TypeScript
• Vite for build tooling
• React Router for navigation
• Tailwind CSS for styling
• Radix UI for accessible components
• Axios for API calls
• Embla Carousel for image carousels

📁 Project Structure

apartments/
├── back/                    # Backend application
│   ├── src/
│   │   ├── config/         # Configuration constants
│   │   ├── controllers/    # Route controllers
│   │   │   ├── apartments.controller.ts
│   │   │   ├── auth.controller.ts
│   │   │   └── embedding.controller.ts
│   │   ├── middleware/     # Authentication middleware
│   │   ├── models/         # MongoDB models
│   │   ├── routes/         # Express routes
│   │   ├── services/       # Business logic
│   │   ├── types/          # TypeScript type definitions
│   │   └── server.ts       # Entry point
│   ├── package.json
│   └── tsconfig.json
│
└── front/                   # Frontend application
    ├── src/
    │   ├── api/            # API client configuration
    │   ├── components/     # React components
    │   │   ├── ui/         # Reusable UI components
    │   │   ├── ChatWidget.tsx
    │   │   ├── FiltersBar.tsx
    │   │   └── ...
    │   ├── contexts/       # React contexts (ChatContext)
    │   ├── hooks/          # Custom React hooks
    │   ├── pages/          # Page components
    │   ├── types/          # TypeScript type definitions
    │   └── main.tsx        # Entry point
    ├── package.json
    └── vite.config.ts

📦 Installation

Prerequisites

• Node.js (v18 or higher)
• MongoDB (local or cloud instance)
• Ollama (for AI chat functionality - gemma2 model)

Setup Steps

1. Clone the repository

   git clone https://github.com/BarYosef212/Apartments-AI.git
   cd Apartments-AI

2. Install backend dependencies

   cd back
   npm install

3. Install frontend dependencies

   cd ../front
   npm install

4. Set up environment variables

   Create a .env file in the back/ directory:

   PORT=4000
   MONGO_URI=your_mongodb_connection_string
   SESSION_SECRET=your_session_secret_key
   JWT_SECRET=your_jwt_secret_key

5. Install and configure Ollama (for AI chat)

   # Install Ollama from https://ollama.ai
   # Then pull the gemma2 model:
   ollama pull gemma2

🚀 Running the Project

Development Mode

1. Start the backend server

   cd back
   npm run dev

   The backend will run on http://localhost:4000

2. Start the frontend development server

   cd front
   npm run dev

   The frontend will run on http://localhost:5173

Production Build

1. Build the backend

   cd back
   npm run build
   npm start

2. Build the frontend

   cd front
   npm run build
   npm run preview

🔌 API Endpoints

Authentication

• POST /auth/login - User login
• POST /auth/register - User registration (if implemented)

Apartments

• GET /apartments - Get paginated list of apartments with filters
  • Query params: page, city, min_price, max_price, rooms, type, search
• GET /apartments/:id - Get apartment by ID
• GET /apartments/cities - Get list of all cities
• GET /apartments/types - Get list of all property types
• POST /apartments - Create new apartment (admin only)
• PUT /apartments/:id - Update apartment (admin only)
• DELETE /apartments/:id - Delete apartment (admin only)

Chat/Embeddings

• POST /embedding/extract-filters - Extract search filters from natural language
• POST /embedding/talk - Chat with AI assistant

🎯 Key Features Explained

AI Chat Interface

The chat widget uses Ollama's gemma2 model to:

• Understand natural language queries in Hebrew
• Extract search filters (city, price range) from conversations
• Provide friendly, conversational responses
• Switch between search mode and chat mode

Semantic Search

Properties are embedded using the Xenova/multilingual-e5-base model:

• Each apartment is converted to a searchable text representation
• User queries are embedded and matched against property embeddings
• Enables intelligent, context-aware property matching

Admin Dashboard

Protected route (/admin) that requires authentication:

• View all apartments in a table format
• Create, edit, and delete apartment listings
• Manage apartment data with a user-friendly interface

🔒 Authentication

The application uses JWT tokens for authentication:

• Tokens are stored in HTTP-only cookies or localStorage
• Protected routes require valid authentication
• Session management via express-session

🌐 Supported Languages

• Hebrew (עברית): Primary language for UI and AI chat
• The AI assistant is configured to respond in Hebrew with a friendly, conversational tone

📝 Environment Variables

Backend (.env)

• PORT - Server port (default: 4000)
• MONGO_URI - MongoDB connection string
• SESSION_SECRET - Secret for express-session
• JWT_SECRET - Secret for JWT token signing

🧪 Development Notes

• The backend uses TypeScript with strict type checking
• Frontend uses Vite for fast HMR (Hot Module Replacement)
• Both projects have separate tsconfig.json files
• UI components are built with Radix UI for accessibility

---

Note: Make sure Ollama is running and the gemma2 model is available before using the chat functionality. The embedding models will be downloaded automatically on first use.