Chatbot SaaS - Retrieval-Augmented Generation (RAG) Application

A chatbot application leveraging a Retrieval-Augmented Generation (RAG) system, built with a microservices architecture using FastAPI, Docker Compose, and Kubernetes. This project integrates a robust architecture to deliver high-quality, responsive conversational AI.

---

Architecture Overview

The system is divided into six primary containers to handle distinct responsibilities effectively:

1. Database Service (Postgres):

   • Uses PostgreSQL with the pgVector extension for storing vector data, enabling document retrieval and context reuse.

2. Redis Service:

   • A short-term database that caches conversations to enhance the chatbot's response quality by maintaining context.

3. LLM Service:

   • Hosts the Large Language Model (e.g., Meta's Ollama) to generate responses in-house, reducing reliance on external LLM providers.

4. Retrieval Service:

   • Manages text transformation, embedding generation, and document retrieval. Connects with the database to ensure data safety and adds a layer of protection.

5. AI Service:

   • Processes queries using relevant documents retrieved by the retrieval service, and communicates with the LLM model to generate final responses.

6. Orchestrator Service:

   • Serves as the API gateway, ensuring seamless integration between services and providing clean public-facing endpoints.

7. UI Service:

   • A web application enabling users to upload text files and pose questions. Built to provide a user-friendly interface for interacting with the chatbot.

---

Features

• Microservices Architecture: Ensures modularity and scalability.
• Dockerized Services: Simplifies deployment and testing across environments.
• RAG System Integration: Combines document retrieval with generative AI for contextually accurate responses.
• LLM Deployment: Supports in-house LLM hosting for cost-efficiency.
• Interactive Web UI: Provides an easy-to-use platform for users to interact with the system.

---

Setup

To get started with the project:

1. Ensure you have Docker and Docker Compose installed on your system.

2. Clone the repository:

   git clone mahdjourOussama/Chatbot
   cd mahdjourOussama/Chatbot

3. Start the services using Docker Compose:

   docker-compose up --build

4. For Kubernetes deployment, apply the configurations:

   kubectl apply -f ./k8s

---

Folder Structure

./
├── services/
│   ├── orchestrator/
│   ├── ai/
│   ├── retrieval/
│   ├── database/
│   ├── llm/
│   ├── redis/
│   └── ui/
├── k8s/
│   ├── deployment.yml
│   ├── service.yml
├── assets/
│   └── chatbot/
│       └── architecture_diagram.png
└── README.md
└── .env
└── docker-compose.yaml

---

Technologies Used

• Backend Framework: FastAPI
• Frontend Framework: Streamlit
• Database: PostgreSQL with pgVector extension
• Cache: Redis
• Orchestration: Docker Compose, Kubernetes
• AI Frameworks: LangChain, Meta Ollama

---

Screenshots

Below are some screenshots from the project:

• Building the images
• Containers Created
• Services Logs
• Docker Desktop
• Main Application
• Start Chat with exesting Document
• Chat interface
• Chat Loading response
• Upload Document
• Document Uploaded

Demo

Check out our live demo video to see the Chatbot SaaS in action:

Click the image above to watch the demo video.

---

For questions or support, please reach out via LinkedIn