<p>Using soft computing (SC) techniques, mining engineers can increase the accuracy and reliability of slope stability predictions. This paper provides a comprehensive framework for developing an optimum predictive model to assess the stability of waste dumps using artificial neural networks. SC approaches are used to evaluate dump stability with factor of safety (FoS). Cross-validation was utilized to evaluate the model’s accuracy with original datasets partitioned into test, validation, and training sets. The performance metrics R² (coefficient of determination), demonstrated good predictive power, with a value of 0.9504 for the total dataset. With R² = 0.9542 for a given dataset subset, the ANN MLP 8-9-1 architecture performed better than others. The study used correlation matrix analysis and MATLAB to determine the relative relevance of input variables, which increased the interpretability and applicability of the developed models. This method not only confirms the model reliability to predict outcomes, also identifies the key variables influencing dump stability. These models allow for proactively measures to reduce hazards associated with dump slope failures, hence enhancing operational efficiency and safety in mining operations through reliable assessment of stability concerns.</p>