This paper presents an integrated Driver Safety Assistant System designed for heavy-duty commercial vehicles in India. The system combines four safety modules: sleep detection using Eye Aspect Ratio (EAR) monitoring with OpenCV, blind spot monitoring using multi-camera feeds, traffic sign recognition using YOLOv5 deep learning model (achieving 93.7% mAP), and an automated Emergency SOS system with GPS-based alert dispatch. The system is deployed on NVIDIA Jetson Nano embedded hardware and targets the critical road safety gap in India's commercial vehicle fleet. Experimental results demonstrate real-time performance with sleep detection response under 3 seconds and traffic sign inference at 28 FPS.