<p>The increasing emphasis on the sustainable construction materials has deepened research on alkali-activated geopolymer concrete (GPC); however, its complex reaction mechanisms and nonlinear mix–property relationships make reliable strength prediction challenging. In the present study, a machine learning (ML) based framework has been developed in order to predict ambient- and heat-cured compressive strength of alkali-activated GPC using data obtained from laboratory evaluation of concrete specimens. Seven different input and two output parameters, have been considered in the current study. The dataset was divided using an 80:20 training-to-testing ratio, and four ML models, namely linear regression (LR), random forest (RF), support vector machine (SVM), and multilayer perceptron (MLP), were implemented. Additionally, the performance of ML models has been assessed using various error metrics, coefficient of determination, regression error characteristic curve, Q-Q plots and K-fold cross-validation. The output of the study indicated that the RF achieved the highest prediction accuracy for both curing conditions (<i>R</i>² ≤ 0.87), whereas the LR, SVM, and MLP produced moderate to poor performance with lower accuracy and greater error values. Further, the feature analysis suggested that input parameters such as binder content, alkali-activated slag content, the ratios of alkali-activated slag to binder content and the NaOH: Na₂SiO₃ were the most influential parameters in affecting the output parameters among others, heat- and ambient-cured strength. To cater to diverse practices, a graphical user interface has been developed to provide instant results that support decision-making and mix design optimization of GPC prior to time- and resource-intensive work.</p>