<p>Accurately predicting concrete’s compressive strength (f_c) is crucial for assessing the quality of in-situ concrete without resorting to destructive testing (DTs). In order to predict the compressive strength of concrete, current research recommends applying a variety of hyper-tuned machine learning algorithms by grid search&#xa0;cross validation when combined with non-destructive testing procedures. The primary objective of this research is to assess the compressive strength of ongoing construction using non-destructive testing methods (NDTs), such as the rebound hammer (RH) and ultrasonic pulse velocity (UPV). Additionally, the study aims to enable field engineers to rapidly evaluate various concrete properties, as the dataset encompasses a wide range of variables. In the methodology of the current research ten different machine learning (ML) techniques were explored using data gathered from diverse literature sources. Subsequently, a tuning process, employing grid search cross-validation (CV), was conducted on the top five ML models. The use of grid search cross-validation techniques greatly improved prediction accuracy and reduced modelling errors that is RMSE (Root Mean Square Value) value across all five machine learning models. Initially the evaluation showed that random forest regression outperformed other models. Further, the R² value of the Random Forest Regressor (RFR) has improved from 0.9 to 0.93 following hyperparameter tuning using Grid Search in the machine learning model. Using the grid search cross validation function, the research has developed and validated hyper-tuned machine learning models, with Random Forest Regression proving to be the most successful. The ultimate aim is to use these models for on-site assessment of the current structure’s strength based on NDTs and their corresponding mixture compositions.</p>