M1 Max 컴퓨터와 로컬 ML 모델로 669GB GoPro 영상을 인덱싱함

This project demonstrates that large-scale video indexing (669GB) is now practical on consumer hardware using local open-source AI models, eliminating cloud dependencies and privacy concerns. The multi-modal vector search capability—combining transcripts, faces, visuals, and audio—shows how rapidly evolving open-source models are enabling sophisticated personal media workflows. This marks a significant shift toward on-device intelligence, where individuals can process sensitive data locally without compromising privacy.