Manual attendance systems and the absence of personalised study guidance remain persistent chal-lenges in modern academic environments. This paper presents StudySphere AI, a lightweight, browser-based academic management system that addresses both problems through two integrated AI subsystems:

(i)   a two-stage contactless authentication pipeline combining Optical Character Recognition (OCR) for in-stitutional ID verification with real-time, client-side face recognition for automated attendance marking, and

(ii)   a feedback-driven adaptive study-recommendation engine that ranks subjects using a weighted priority score derived from attendance ratio, consecutive absences, study recency, and self-reported focus level. The system is built on React 18, TypeScript, and a PostgreSQL/Supabase backend. All biometric inference runs entirely in the browser via face-api.js, ensuring no video data leaves the user’s device. Experimental eval-uation confirms reliable face detection (confidence ≥ 60%) under standard indoor lighting, accurate OCR authentication (similarity ≥ 80%), and correct adaptive priority-based subject ranking. A faculty analytics dashboard provides cohort-level attendance insights and at-risk student identification. The system demon-strates that a single, hardware-free web application can replicate functionality previously requiring dedicated biometric devices and complex learning management systems.

Keywords: Artificial Intelligence, Face Recognition, OCR, Attendance Automation, Adaptive Learning, Study Recommendation, Academic Analytics, Privacy-Preserving AI.