Effective supply chain management (SCM) is crucial for improving efficiency, reducing costs, and enhancing responsiveness in dynamic market conditions. Traditional SCM optimization methods often rely on static models that struggle to adapt to uncertainty and real-time changes. In this study, we propose a data-driven approach using reinforcement learning (RL) to optimize decision-making in SCM. By leveraging historical and real-time data, our RL model dynamically learns optimal inventory policies, demand forecasting strategies, and logistics planning to minimize costs and maximize service levels. We evaluate the performance of our approach through simulations and real-world case studies, demonstrating significant improvements over conventional optimization techniques. The results highlight the potential of RL in transforming SCM by enabling adaptive, intelligent decision-making in complex and uncertain environments.