RapidSummarize 📄⚡

RapidSummarize is a cloud-native Retrieval-Augmented Generation (RAG) application that lets you upload PDFs and have intelligent conversations with your documents.

Built with a React frontend and FastAPI backend, powered by AWS Bedrock (Meta Llama 4 Scout + Amazon Titan Embeddings V2), with ChromaDB for vector storage and Amazon S3 for document storage — deployed on AWS EC2 with Docker.

🌿 Looking for the local version? See the main branch (runs entirely with Ollama, no cloud required).

---

🚀 Live Demo: Available on request — contact me to spin up the instance.

✨ Features

• 📁 Upload multiple PDFs at once (up to 100 pages, 10MB per file)
• 🔍 Intelligent text extraction with OCR support for scanned pages
• 🧠 Context-aware conversations with follow-up question support
• ⚡ Real-time streaming responses via AWS Bedrock
• 🗂️ File management: select, delete, clear context
• 🔐 Session-based isolation — each user sees only their own documents
• 🐳 Fully containerized with Docker Compose
• ☁️ Cloud-hosted on AWS EC2 with S3 document storage

---

🛠️ Built With

Frontend

• React + Vite
• Nginx

Backend

• FastAPI
• PyMuPDF + pytesseract (OCR)
• ChromaDB (vector storage)
• boto3 (AWS SDK)
• slowapi (rate limiting)

AI & Cloud

• LLM: Meta Llama 4 Scout 17B (via AWS Bedrock)
• Embeddings: Amazon Titan Text Embeddings V2 (via AWS Bedrock)
• Document Storage: Amazon S3
• Hosting: AWS EC2 (Ubuntu, Docker Compose)
• Auth: AWS IAM roles (no hardcoded credentials)

Containerization

• Docker
• Docker Compose

---

🏗️ Architecture

User Browser
    │
    ▼
EC2 Instance
├── Nginx (port 80) → React Frontend
├── FastAPI Backend (port 8000)
│   ├── S3 → PDF storage
│   ├── ChromaDB → vector index
│   └── AWS Bedrock → Titan Embeddings + Llama 4 Scout

---

🚀 Deploy Your Own (AWS)

Prerequisites

• AWS account with Bedrock access (Titan Embeddings V2 + Llama 4 Scout)
• EC2 instance (t3.micro or larger)
• S3 bucket
• IAM role with AmazonBedrockFullAccess + AmazonS3FullAccess attached to EC2

1️⃣ Clone the repository

git clone -b aws-bedrock https://github.com/VeeSeven/RapidSummarize.git
cd RapidSummarize

2️⃣ Configure docker-compose.yml

environment:
  - ALLOWED_ORIGINS=http://your-ec2-ip
  - AWS_DEFAULT_REGION=us-east-1
  - S3_BUCKET=your-bucket-name

args:
  - VITE_API_URL=http://your-ec2-ip:8000

3️⃣ Start the application

docker compose up -d --build

4️⃣ Open the app

http://your-ec2-ip

---

📖 How to Use

1. Upload PDFs — select one or more PDFs on the home page, optionally enter an initial question
2. Wait for processing — the app automatically waits for PDF indexing to complete before sending your first query
3. Chat with your documents — answers stream in real time, based only on your selected PDFs
4. Manage files — delete PDFs, clear chat history, or switch document context anytime

---

🔌 API Overview

Method	Endpoint	Description
POST	/upload	Upload PDFs (multipart/form-data)
GET	/files	List session PDFs
DELETE	/files/{filename}	Delete a PDF and its vectors
GET	/status/{filename}	Check PDF processing status
POST	/chat	Query selected PDFs
POST	/chat-with-context	Query with previous chat context

All responses are JSON or streaming text. Endpoints are rate limited (10 req/min for chat, 5 req/min for upload).

---

⚙️ Configuration

Variable	Description
ALLOWED_ORIGINS	CORS origins (comma-separated)
AWS_DEFAULT_REGION	AWS region (default: us-east-1)
S3_BUCKET	S3 bucket name for PDF storage
VITE_API_URL	Backend API URL (build-time)

---

💾 Storage

Store	Purpose
Amazon S3	Uploaded PDF files (persistent, per-session)
ChromaDB	Vector embeddings for semantic search
EC2 EBS	Temporary processing only

---

🧪 Local Development (Ollama version)

See the main branch for the fully local setup using Ollama — no AWS account needed.

---

Thanks

• Meta & AWS — for Llama 4 Scout via Bedrock
• Amazon — for Titan Embeddings V2
• ChromaDB — for the vector database
• FastAPI — for the backend framework
• React & Vite — for frontend tooling

---

Happy Summarizing!