Predicting the settlement of nodular piles under static loading is challenging due to the nonlinear nature of pile–soil interaction. In this study, we use a hybrid model that combines Artificial Neural Networks (ANNs) with the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) to tune hyperparameters and network architecture automatically. The model is trained using experimental data that include pile geometry, applied load, and soil conditions. To interpret the model, we apply feature importance method. The results show that the ANN–CMA-ES model produces accurate predictions and identifies the most important input variables, such as load and cylindrical diameter. This modeling approach may help improve decision-making in pile foundation design.