Classroom AI Analytics Pipeline ingests video and audio from five camera and microphone sources into an on-premises encrypted archive, then processes streams through specialized ML models for speech-to-text with speaker diarization, emotion and engagement detection, and behavior analysis. An agentic AI layer synthesizes outputs from all three models into unified insights. Four role-specific dashboards—Teaching, Learning, Speech, and STP—feed into a consolidated classroom report, enabling educators to optimize instruction while maintaining AES-256 encryption and zero cloud data exposure. Fork this diagram to customize sensor placement, add additional ML models, or adapt the dashboard schema for your institution's analytics requirements.