Real-time hand gesture recognition system using MediaPipe for slide control and mouse input without physical devices. Video frames from a webcam feed through preprocessing, hand landmark extraction via MediaPipe's 21-point model, and gesture classification using CNN/LSTM to detect swipes, pinches, and finger states. The Gesture Recognition Engine outputs confidence-scored commands to a Control Dispatcher that maps gestures to PowerPoint/Keynote slide navigation and OS-level mouse control via PyAutoGUI. This architecture enables touchless presentation control and accessibility-focused input, reducing reliance on hardware peripherals while maintaining sub-100ms latency through temporal smoothing and confidence thresholding. Fork and customize this diagram on Diagrams.so to adapt gesture mappings, add new classifiers, or integrate alternative hand tracking models.