<p>The overexploitation of natural sand as fine aggregate in concrete has raised environmental concerns, driving the need for sustainable alternatives in concrete production. However, ceramic tile waste, a significant by-product of construction and demolition, presents a promising alternative to fine aggregates while addressing waste management challenges. This study applies Gene Expression Programming (GEP) to predict the compressive strength of ceramic tile waste-modified concrete. A dataset of 136 unique concrete mixes, compiled from published literature, was used to develop and validate multiple GEP models under varying genetic configurations. The input variables comprise cement (kg/m³), fine natural aggregate, coarse natural aggregate, ceramic fine aggregate (kg/m³), curing time (days), superplasticizer (kg/m³), water (kg/m³), and water-cement ratio. The target output variable is compressive strength (MPa) recorded for each mix at specific curing ages. The dataset was randomly split into two subsets: a training subset (70%) and a testing/validation subset (30%). The coefficient of determination (R), root mean square error (RMSE), mean absolute error (MAE), and relative root mean square error (RRMSE) were used to evaluate the model performance. Among twenty trial models, GEP10 achieved the best performance with R = 0.916 (training) and R = 0.933 (validation), demonstrating strong predictive accuracy. The model achieves RMSE of 4.723&#xa0;MPa and MAE of 3.713&#xa0;MPa. Furthermore, Shapley Additive Explanations (SHAP) were employed to enhance model interpretability, revealing that curing age and water content were the most influential features. At the same time, aggregates and admixtures had minimal impact on the predictive capabilities. The results highlight the robustness of GEP in modeling nonlinear behaviour and provide a reliable predictive framework for sustainable concrete incorporating ceramic tile waste.</p>