In the era of smart cities, efficient urban traffic management has become a critical challenge. This paper suggests an optimized intelligent traffic management system that employs advanced machine learning algorithms to predict and alleviate traffic congestion. The system analyzes high-resolution vehicular network data and provides real-time traffic updates by integrating Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, Gradient Boosting Machines (GBMs), and Transformer models. The Traffic Prediction Dataset is employed to validate the proposed approach, demonstrating superior accuracy and efficiency compared to conventional methods. The results of our study indicate that these advanced machine learning techniques obtain a prediction accuracy of 99.37%, thereby reducing the mean absolute error by 15% compared to the current solutions. Moreover, the system exhibits a 20% enhancement in traffic flow optimization. In smart urban environments, these findings underscore the scalability and robustness of our approach, which provides a dependable solution for traffic management. This research aims to improve the quality of life for city residents, reduce economic loss, and reduce pollution levels. It also contributes to the development of intelligent transport systems.