🚀 PanScience Context-Aware RAG AI

A production-ready, full-stack Retrieval-Augmented Generation (RAG) system with context-aware conversations. Upload documents (PDF/DOCX/TXT) and chat with an AI that remembers your conversation history.

📋 Table of Contents

• Features
• Architecture
• Quick Start
• Running the Application
• Project Structure
• API Documentation

✨ Features

• 📄 Multi-Format Support: PDF, DOCX, TXT, MD files
• 🧠 Context-Aware AI: Remembers last 6 conversation exchanges
• 🔍 Vector Search: ChromaDB-powered semantic retrieval
• 🎯 Intelligent Chunking: 1500-char chunks with 200-char overlap
• 📊 Document Management: Track indexed documents
• 🔒 Production Ready: Health checks, rate limiting, error handling
• 🐳 Fully Dockerized: One-command deployment
• 🌐 Cloud Ready: AWS/GCP/Azure deployment configurations

🏗️ Architecture

         ┌──────────────────────────┐
         │        Frontend          │
         │  React + Tailwind (Vite) │
         └─────────────┬────────────┘
                       │
                       ▼
         ┌──────────────────────────┐
         │        FastAPI API       │
         │  (Document & Chat APIs)  │
         └─────────────┬────────────┘
                       │
        ┌──────────────┼──────────────┐
        ▼              ▼              ▼
┌─────────────┐  ┌──────────────┐  ┌──────────────┐
│  MongoDB     │  │  ChromaDB     │  │  Gemini/OpenAI│
│  (metadata)  │  │  (embeddings) │  │  (generation) │
└─────────────┘  └──────────────┘  └──────────────┘

🚀 Quick Start In Just 2 Steps

Prerequisites

• Docker & Docker Compose (recommended)
• OR Python 3.12+ & Node.js 20+
• Google Gemini API Key (Get one here)

1️⃣ Clone Repository

git clone https://github.com/satvikmishra44/PanScienceRAG.git
cd PanScienceRAG

---

2️⃣ Build & Start Containers

docker-compose up --build

This launches:

Service	Description	URL
🧠 Backend (FastAPI)	Core API	http://localhost:8000
💬 Frontend (React)	UI Dashboard	http://localhost:5173

Just Go To Frontend localhost port by clicking here (after running the docker command) and use the application yourself. As Simple As That

3️⃣ Stop Containers

docker-compose down

---

4️⃣ Verify Setup

Check API health:

curl http://localhost:8000/ping

Expected response:

{"status": "ok", "service": "RAG-pipeline"}

🔗 API Usage Guide

🧩 Base URL

http://localhost:8000

---

📄 1. Upload Document

Endpoint: /ingest
Method: POST

📝 Description

Uploads and indexes a document for retrieval and querying.

💡 Example (cURL)

curl -X POST "http://localhost:8000/ingest" \
  -F "file=@research_paper.pdf"

✅ Response

{
  "status": "success",
  "doc_id": "66e432f78df123abc",
  "chunks": 98
}

---

🔍 2. Query Documents

Endpoint: /query
Method: POST

📝 Description

Ask questions based on the indexed documents.
Supports chat history for contextual and conversational responses.

💡 Example (JSON Body)

{
  "query": "What are the applications of quantum entanglement?",
  "top_k": 4,
  "history": [
    {"role": "user", "text": "Tell me about quantum mechanics"},
    {"role": "ai", "text": "Quantum mechanics studies the behavior of matter and energy..."}
  ]
}

✅ Response

{
  "status": "success",
  "answer": "Quantum entanglement enables applications in quantum computing, teleportation, and cryptography...",
  "sources": [
    {
      "text": "Quantum entanglement is a phenomenon...",
      "meta": {"source_filename": "quantum_intro.pdf"},
      "distance": 0.12
    }
  ]
}

---

📚 3. Get Document List

Endpoint: /documents
Method: GET

💡 Example

curl http://localhost:8000/documents

---

❤️ 4. Health Check

Endpoint: /ping
Method: GET

💡 Example

curl http://localhost:8000/ping

✅ Response

{"status": "ok", "service": "RAG-pipeline"}

---

Project Structure

PanScience/ │ ├── backend/ │ ├── app/ │ │ ├── main.py # FastAPI entry point │ │ ├── db.py # Database + LLM initialization │ │ ├── rag.py # RAG ingestion and query logic │ │ ├── utils.py # File handling and text chunking │ ├── requirements.txt │ ├── Dockerfile │ ├── frontend/ │ ├── src/ │ │ ├── components/ │ │ │ ├── Chat.jsx │ │ │ ├── DocManager.jsx │ │ │ ├── Landing.jsx │ ├── package.json │ ├── Dockerfile │ ├── docker-compose.yaml └── README.md

🐳 Running with Docker

1️⃣ Build & Start Containers

docker-compose up --build

This launches:

Service	Description	URL
🧠 Backend (FastAPI)	Core API	http://localhost:8000
💬 Frontend (React)	UI Dashboard	http://localhost:5173

2️⃣ Stop Containers

docker-compose down

---

🧰 Tech Stack

Layer	Technology
🖥️ Frontend	React (Vite), TailwindCSS
⚙️ Backend	FastAPI, Python
🗃️ Database	MongoDB
🧮 Vector Store	ChromaDB
🧠 LLM	Gemini / OpenAI / Claude
🐳 Containerization	Docker, Docker Compose

---

🧾 Notes

• Ensure MongoDB and ChromaDB services are running before ingesting documents.
• Large PDFs are automatically chunked for efficient retrieval.