Floods are common and destructive natural hazards that require effective management and mitigation strategies. In this study we used two ensemble machine learning (ML) models, namely eXtreme Gradient Boosting (XGB) and Random Forest (RF) to map flood hazard probability in the city of Zaio. Nine flood conditioning factors were employed as input variables for XGB and RF. Historical flood presence and absence data were split into 70% and 30% samples to train and test the models, respectively. The output of our study is a map that displays urban flood hazard in Zaio city. The performance of the two models was assessed through performance metrics, particularly Area under the curve (AUC), accuracy, specificity, and sensitivity. Results emphasize that both XGB and RF showed excellent robustness in flood hazard classification. The findings of our study explore the potential of ensemble ML models in producing reliable urban flood hazard maps, providing an alternative to the computationally expensive hydrodynamic models.