🤖 Ebla Assistant — RAG+SK-Powered Chat App

This project is a chat assistant built for Ebla Computer Consultancy, designed to provide helpful responses using Retrieval-Augmented Generation (RAG). It integrates:

• FastAPI for the backend API
• Ollama model for LLM responses
• MongoDB for chat session storage
• Chroma for vector database
• Custom UI for the chatbot interface

The assistant remembers previous messages in each chat session and uses RAG to provide context-aware responses by retrieving relevant knowledge chunks.

---

🎯 Purpose

This project was created to gain a practical understanding of how Retrieval-Augmented Generation (RAG) works. It demonstrates how RAG can improve the accuracy and relevance of AI responses by combining local context with external vectorized knowledge.

---

🖥️ Features

• Clean chatbot UI for interaction
• Session-based chat history with memory (stored in MongoDB)
• Vector search integration using Chroma
• Ollama model for natural language responses
• Built with modular FastAPI backend

---

📦 Requirements

Make sure you have the following installed:

• Ollama
• Python 3.10+
• pip package manager
• virtualenv (recommended)

---

⚙️ Setup Instructions

1. Clone the Repository

git clone https://github.com/3bdop/chat-app.git
cd chat-app

2. Install Ollama 🦙

Ollama lets you run large language models locally. Follow the instructions for your OS:

• macOS / Linux / Windows:
  https://ollama.com/download

• After installation, verify it's working:

  ollama --version

• Pull llama3 model:

  ollama pull llama3

• Pull mxbai-embed-large model for creating embeddings:

  ollama pull mxbai-embed-large

• Verify installation:

  ollama list

3. FastAPI setup

• Create environment (recommended):

  python -m venv .venv

• Activate environment:

  source .venv/bin/activate    # Windows: .\.venv\Scripts\activate.bat

• Install requirements:

  pip install -r requirements.txt

4. Environment variable setup

• Create a .env file in the root and the following with your own values:

  MONGODB_URI=<your db string>
  MONGODB_DB_NAME=<your db name>

5. Run the chat app

• To run the app type the following:

  fastapi dev src/main.py