🛡️ AI AWS IAM Sentinel

AI AWS IAM Sentinel is an enterprise-grade GenAI assistant designed to provide precise, context-aware answers strictly from the AWS IAM User Guide. It leverages a robust RAG (Retrieval-Augmented Generation) pipeline to ensure high accuracy and security-focused responses.

Live Demo: https://ritesh-iam-sentinel.vercel.app/

---

🚀 Features

• Context-Aware Q&A: Answers questions based solely on the official AWS IAM User Guide.
• Automated Data Pipeline: Automatically parses, chunks, and indexes the AWS IAM PDF guide into a vector database on startup.
• Hybrid Search: Combines semantic search (vector) with BM25 keyword matching for optimal retrieval.
• Enterprise-Ready: Built with FastAPI and Next.js, optimized for production deployments.
• Background Indexing: Performs heavy indexing tasks in a background thread to keep the service responsive.

---

🏗️ Architecture

Tech Stack

• Frontend: Next.js (TypeScript, Tailwind CSS, Lucide React)
• Backend: FastAPI (Python 3.11)
• Vector Database: Qdrant Cloud
• LLM: OpenAI GPT-4o (via LangChain)
• Embeddings: BAAI/bge-small-en-v1.5 (running locally on CPU)
• Orchestration: LangChain & LangGraph

System Flow

1. Ingestion: On startup, the system reads iam-ug.pdf, chunks it using recursive character splitting, and uploads embeddings to Qdrant.
2. Retrieval: When a user asks a question, the system retrieves relevant chunks using the BGE embedding model.
3. Generation: The context is passed to GPT-4o with a strict "answer only from context" prompt.

---

🛠️ Local Setup

Prerequisites

• Docker & Docker Compose
• OpenAI API Key
• Qdrant Cloud Account (Free tier works)

Environment Setup

1. Clone the repository.
2. Create a .env file in the root directory (use .env.example as a template):

   cp .env.example .env

3. Fill in your credentials:
   • OPENAI_API_KEY: Your OpenAI key.
   • QDRANT_URL & QDRANT_API_KEY: Your Qdrant Cloud instance details.
   • DATABASE_URL: Your database connection string (if using persistence).

Running with Docker

The easiest way to run the entire system:

docker-compose up --build

• Frontend: http://localhost:3000
• Backend API: http://localhost:8000
• API Docs: http://localhost:8000/docs

---

📦 Deployment

Render / Production

This project is optimized for deployment on Render.

• The Dockerfile includes pre-caching of embedding models to prevent startup timeouts.
• Use docker-compose.prod.yml for production-grade configurations.

Hugging Face Spaces

The project includes support for Hugging Face Spaces:

• See backend/Dockerfile.hf for the specific HF configuration.
• The system is designed to run efficiently on CPU-only environments.

---

📂 Project Structure

├── backend/            # FastAPI Application
│   ├── app/            # Core logic, APIs, and services
│   ├── data/           # PDF source files
│   └── scripts/        # Utility scripts
├── frontend/           # Next.js UI
│   ├── src/            # Components and app logic
│   └── public/         # Static assets
├── data/               # Persistent data volume
└── docker-compose.yml  # Orchestration

---

📄 License

This project is for educational and enterprise demonstration purposes. Data source: AWS IAM User Guide.