The idea of smart cities focuses on leveraging modern technologies to enhance and streamline city operations, particularly energy infrastructure. One of the key challenges that smart cities face is ensuring the efficient management of energy resources to minimize consumption, costs, and environmental impact. Machine Learning (ML) provides a powerful means to optimize energy usage within urban infrastructure. This paper introduces a framework to optimize energy management in smart cities by employing ML techniques. The framework comprises three primary components: data collection, model development, and energy optimization. Data collection entails gathering energy consumption information from various sources like smart meters, sensors, and other Internet of Things (IoT) devices. After data preprocessing and cleaning, ML models, using techniques such as regression, classification, clustering, and deep learning, are developed to forecast energy consumption and optimize usage. The optimization process then utilizes these models to balance energy supply and demand, ultimately reducing overall consumption and cost. The framework is advantageous in decreasing energy use, lowering costs, and reducing environmental impacts while improving the reliability and efficiency of urban energy infrastructure. This solution can be applied across smart city domains such as buildings, transportation, and industrial activities.