Proxy attendance, whereby attendance is recorded for a student who is absent, continues to pose a problem in higher education institutions, since single-factor verification methods like static QR codes, RFID cards, and GPS-only location checks each have a well-documented, low-effort way of being circumvented. Although more recent multi-factor methods that combine QR codes, facial recognition, and GPS-based geofencing have improved on the single-factor approaches, they still depend on satellite-based location verification, which suffers from considerable degradation in accuracy in indoor, multi-storey classroom settings, and generally assess the various factors together using a strict conjunctive policy without taking into account the increased false-rejection rate that this causes for students who are truly present. The present paper puts forward a multi-factor, zero-trust attendance verification framework that replaces GPS-based geofencing with ad-hoc Bluetooth Low Energy (BLE) proximity detection broadcasts sent from the instructor's own smartphone, thus getting rid of both the indoor accuracy problems associated with GPS and the need for dedicated beacon hardware that is required by fixed-infrastructure BLE systems. The framework also combines a quickly rotating cryptographically-signed QR token with an on-device face-liveness check and includes persistent device-hardware binding, and introduces a weighted-confidence fusion policy which ensures strong protection against proxy attacks while at the same time explicitly reducing the compounded false-rejection rate that is inherent in using a strict multi-factor conjunction. The system's architecture, the design of the individual modules, and the verification process are described, together with a probabilistic analysis of the reliability trade-offs involved. Only the authentication and device-binding module have been implemented thus far, and an experimental evaluation of the entire system has yet to be carried out.