OpenHand

OpenHand was special to me because it was the first time I combined machine learning and education. I trained a Random Forest classifier on MediaPipe and OpenCV hand landmarks, and actually looked inside it instead of treating it like a black box. The tricky part was making it run live. The Python backend was processing frames while the Next.js frontend expected instant predictions, and at first it lagged and felt clunky. I had to clean up the socket flow, cut unnecessary data and optimize how the frontend and backend talked to each other. That showed me how much the data flow between the parts of a system matters.