Smart grid integration and deregulated electricity markets have complicated power marketing ecosystems, necessitating flexible solutions to ensure grid stability, economic competitiveness, and operational efficiency through precise forecasting, real-time decision-making, and enhanced service delivery. Conventional forecasting models struggle with accuracy, scalability, and robust optimization, especially when dealing with non-linear, high-dimensional energy datasets. Moreover, they lack adaptability to volatile demand–supply dynamics and real-time operational constraints. This research presents a next-generation intelligent platform that integrates a Weighted Golden Eagle Optimized Light Gradient Boosting Machine (WGEO-LightGBM) for predictive modeling and operational optimization in electricity marketing and service delivery. Data is collected from IoT-based smart meters and supervisory control and data acquisition (SCADA) systems, encompassing load demand, pricing signals, and historical consumption trends. The collected data were preprocessed using missing value imputation, Min-Max scaling, and Recursive Feature Elimination (RFE) to enhance model input quality and reduce feature redundancy. The WGE algorithm enhances the search efficiency and solution diversity by adaptively adjusting flight dynamics and prey targeting mechanisms. LightGBM, a fast, scalable regression model, is used for short-term load forecasting and price prediction, enabling real-time pricing, demand-side management, efficient service delivery, and resource allocation. The WGEO-LightGBM model predicted load values ranging from approximately 120 MW–160 MW across multiple hourly intervals, closely aligning with actual observations and demonstrating its precision in dynamic grid environments. Its integration into intelligent platforms offers scalable, data-driven solutions for real-time energy management and improved service delivery.