Predictive modeling of radioactive waste behavior inside cloud-based systems is presented using a unique technique leveraging dynamic Bayesian Networks (BN). The complicated and ever-changing nature of radioactive waste management makes it very difficult to make informed decisions without the use of advanced forecasting technologies. When trying to simulate the complex behavior of radioactive waste over time, traditional modeling approaches generally fall short. Considering the inherent uncertainty and time interdependence in radioactive waste systems it proposes using BN as a solution. Scalable and efficient processing, made possible by cloud computing resources, is essential for meeting the computational needs of BN. A case study of radioactive waste leaching behavior prediction at a storage facility is used to illustrate the efficacy of the proposed method. The findings show that BN-based technique is more accurate and scalable than conventional modeling approaches, which may be used to improve waste management tactics. Decisions may be made with more knowledge and risk can be reduced advancements in radioactive waste management's predictive modeling skills. © 2024 Elsevier B.V., All rights reserved.