The complete architecture and documentation for EyeDriveSafe are now publicly available. ​ Developed as concluding STEM research, the project addresses road safety by deploying a MobileNetV2 transfer learning model on an Arduino Uno Q for real-time edge AI facial analysis. The system utilizes whole-frame binary classification to compute continuous drowsiness metrics and translates visual cues into an escalating three-tier alert state machine. ​ The formal research paper is...