---

VIDEO DEMO

https://www.youtube.com/watch?v=hXF-gVWQTIg

📌 AI Chatbot for Customer Support

This project is an AI-powered chatbot that provides customer support by answering questions about a judicial circular. The chatbot is designed with a microservices architecture, allowing modular development and scalability. It leverages Retrieval-Augmented Generation (RAG) to retrieve relevant information and generate context-aware responses.

🚀 Services

The project consists of the following services:

1. Frontend: A React application styled with TailwindCSS, providing an interactive user interface to chat with the AI-powered assistant.

2. Service2 (API Gateway): A FastAPI-based backend that acts as a communication bridge between the frontend and Service3. It also manages user interaction history.

3. Service3 (Chatbot Engine): Another FastAPI-based backend hosting the chatbot logic. It uses LangChain and OpenAI’s API to retrieve information from the circular and generate accurate responses.

4. Redis: A Redis server used as a vector database for state management and caching.

5. Postgres (pgvector): A PostgreSQL server with pgvector extension, used for storing embeddings of the circular’s text to optimize retrieval operations.

🏗️ Microservices Architecture

This project follows a microservices-based architecture, which simulates a real-world software development environment. Each service is containerized using Docker for improved portability and deployment efficiency.

Key Decisions

• Technology Selection: React, FastAPI, Redis, pgvector, LangChain, and Docker were chosen for their efficiency and compatibility.
• Chunk Size Optimization: The judicial circular is split into optimized chunks to improve retrieval accuracy in the RAG pipeline.
• Microservices Implementation: This ensures modularity, scalability, and better collaboration in a team setting.
• Vector Databases: Using Redis and pgvector significantly improves query response time by leveraging efficient embedding-based searches.
• Docker for Portability: All services are containerized, allowing seamless deployment across different environments.

🔑 Environment Variables

Rename the .env.example file to .env and configure your OpenAI API Key before running the application.

🛠️ Setup

To run the project, ensure you have Docker installed on your system.

Then, execute the following command:

docker-compose up --build

This will start all services as defined in docker-compose.yml, pulling necessary Docker images, creating containers, and running them together.

📥 Data Population

The database must be populated with embeddings from the judicial circular before the chatbot can function correctly.

1️⃣ Set up a virtual environment (recommended)

It is recommended to use Anaconda to manage dependencies.

python -m venv venv
.\venv\Scripts\activate

2️⃣ Install dependencies

Ensure that all required libraries are installed before running the script:

pip install -r requirements.txt

3️⃣ Run the population script

Once the virtual environment is active and dependencies are installed, run the script:

python insert_data.py

This script will:
✔ Connect to the PostgreSQL (pgvector) database
✔ Create necessary tables (if they don’t exist)
✔ Insert the processed embeddings into the vector database

---