In Saudi Arabia, the development of smart cities is leading to the development of transportation systems and infrastructure that include intelligent transportation, energy consumption, and public services. The current review paper explores the possibility of applying machine learning methods for traffic flow predictions and energy optimization in smart Saudi cities. For this purpose, the paper will use the systematic review methodology, which will analyze the peer-reviewed articles published between 2020 and 2025, relevant policy papers, and real-life applications related to Saudi smart cities. In contrast to the existing literature, which usually explores separate aspects of traffic analytics or energy management, the proposed research will focus on integrating two areas using urban optimization approach. The review reveals that recurrent neural networks, long short-term memory models, graph neural networks, ensemble and reinforcement learning are highly promising methods that can be used for predicting congestion, coordinating signals, controlling the process of electric-vehicle charging, managing street lighting and minimizing the waste of energy in cities. Successful application of machine learning, however, will depend on various conditions such as data quality, cybersecurity issues, interoperation, and proper governance.