With the proliferation of vehicle-to-everything (V2X) technologies, a number of novel use cases for enhanced road safety are emerging. A typical application utilizing this technology is the protection of vulnerable road participants such as pedestrians or cyclists. In this dissertation, a real-time notification system is presented to enhance and improve the safety and user experience of vulnerable road participants. The proposed technology utilizes hybrid communications capabilities (among road participants and between road participants and edge computers), advanced machine learning based movement prediction, and edge computing principles to deliver a cross-platform solution with real-time efficiency. To achieve this goal, this research work focuses on solving three major challenges: First, utilize multiple wireless communication technologies in a hybrid network environment to allow data sharing among different hardware platforms and user groups; Second, effectively identify and predict high-risk traffic events using advanced machine learning techniques based on historical data and user profiles; Finally, leverage state-of-the-art edge computing framework to efficiently distribute the computing workload and minimize latencies. The performance of the proposed system is evaluated using simulation methods as well as case studies on the real-world testbed.