<p>Fog computing has emerged as an effective paradigm to address the latency issues inherent in cloud computing by enabling data processing closer to data sources, particularly for IoT applications in healthcare, smart transportation, and industrial automation. However, efficient task scheduling in fog environments remains challenging due to their dynamic, heterogeneous nature. This paper presents an integrated solution for task scheduling in fog-cloud environments to reduce energy consumption and minimize task makespan, while maintaining high-quality service (QoS). The proposed model comprises three core components: workload prediction, task offloading, and scheduling optimization. First, a Feedforward Neural Network (FNN) is used to predict workload, allowing for proactive resource allocation. Next, a reinforcement learning (RL) model dynamically manages task offloading decisions based on real-time performance metrics, determining whether tasks should be processed on local fog nodes or offloaded to the cloud. Finally, a hybrid optimization algorithm combining Harris Hawk Optimization (HHO) and Genetic Algorithm (GA) refines task scheduling by mapping tasks to available resources efficiently. Simulation results show that the proposed method achieves significant improvements in energy efficiency, reduces task completion times, and enhances QoS metrics compared to baseline scheduling algorithms. The novelty of this work lies in the synergistic integration of predictive deep learning, adaptive RL-based offloading with a multi-objective reward function, and a hybrid HHO-GA optimizer featuring a customizable four-objective fitness function, which collectively address dynamic workload fluctuations and multi-objective trade-offs unmet by existing hybrid methods.</p>