Developed an AI-driven attendance system using YOLOv8 & FaceNet, integrating a Vector Database ChromaDB for efficient face embedding storage and retrieval. Eliminated the need for retraining by leveraging retrieval-augmented generation (RAG)-style lookups, reducing computational overhead and improving accuracy. 1. Face Embedding Storage: Used ChromaDB to store and retrieve face embeddings, enabling instant recognition without retraining. 2. Real-Time Attendance Marking: Captured classroom images and matched them against stored embeddings for automated search, filter, and verification. 3. FastAPI Deployment: Deployed the system using FastAPI, ensuring high-speed inference and scalable API integration for seamless attendance tracking. 4. Scalable Web Application: Built with FastAPI & Bootstrap, providing student management, attendance history, and cross-device accessibility. 5. Enhanced RAG-based Face Retrieval: Improved system efficiency using approximate nearest neighbor (ANN) search, enabling fast, scalable face recognition.