In the digital era, the mining industry benefits from powerful tools that can help to optimise underground backfilling operations and to increase overall safety. Indeed, with current progress in artificial intelligence (AI), machine learning (ML) creates state-of-the-art techniques in the mining sector that could significantly improve the productivity and efficiency of mining operations. The purpose of this study is to apply ML algorithms, including the gradient boosting regressor (GBR), the XGBoost regressor (XGBR), and the support vector regressor (SVR) to predict the uniaxial compressive strength (UCS) of cemented paste backfill (CPB). A total of 1,587 UCS data were used to train the ML algorithms, considering different variables such as the types of tailings, binder and their proportion, solid mass concentration, slump height, water quality, and curing time. The raw data were pre-processed before training the models, as well as their hyperparameters tuning was made by a random search method followed by 4-fold cross-validation. The prediction results show that the GBR algorithm is the most powerful one which has a coefficient of correlation (R) between predicted and experimental values equal to 0.99 and a root-mean-square error (RMSE) equal to 0.16. This prediction is validated through new-lab prepared CPB specimens.