🧠 AI RAG Chatbot Backend

Memory • Retrieval-Augmented Generation (RAG) • FastAPI • LangChain • Vector DB

---

📌 Project Overview

This project implements an enterprise-grade backend for a Conversational AI Chatbot using FastAPI and LangChain.

The chatbot supports:

• ✅ Session-based conversational memory
• ✅ Retrieval Augmented Generation (RAG) using PDFs
• ✅ Vector search using ChromaDB
• ✅ Source citations with answers
• ✅ Extensible architecture (Auth, Streaming, Tools ready)

---

🏗️ Architecture

Client → FastAPI → LangChain (Memory + RAG) → ChromaDB → Documents

---

🧰 Tech Stack

• FastAPI
• LangChain (classic + openai)
• OpenAI
• ChromaDB
• PyPDF
• Uvicorn
• Python 3.10+

---

📁 Folder Structure

ai_rag_chatbot_backend/
├── app.py
├── config.py
├── database.py
├── models.py
├── auth/
├── memory/
├── rag/
├── tools/
├── analytics/
├── streaming/
├── utils/
├── data/
├── vectorstore/
├── .env
└── README.md

---

🔐 Environment Variables

Create a .env file:

OPENAI_API_KEY=your_openai_api_key_here
CHROMA_PERSIST_DIR=vectorstore

---

▶️ Step-by-Step Setup

1️⃣ Create Virtual Environment

python -m venv venv
venv\Scripts\activate

2️⃣ Install Dependencies

pip install fastapi uvicorn pydantic python-dotenv langchain langchain-classic langchain-openai langchain-community chromadb pypdf

3️⃣ Add PDFs

Place PDFs inside:

data/

4️⃣ Ingest Documents

python rag/ingest.py

5️⃣ Run Server

uvicorn app:app --reload

Open:

http://127.0.0.1:8000/docs

---

🧪 Test API

POST /chat

{
  "message": "What is RAG?",
  "session_id": "test-1",
  "memory_type": "buffer"
}

---