In the FinTech era, the evolution of credit scoring systems is being rapidly accelerated by advances in machine learning and data analytics. Accurate credit prediction is vital for financial institutions to minimize risk and streamline lending decisions. With the expansion of digital lending platforms and alternative credit channels, there is an urgent need for intelligent, data-driven models that can evaluate creditworthiness in real time. This study proposes a credit scoring prediction model using the XGBoost algorithm, which demonstrates exceptional classification performance in identifying good and bad credit customers. The model is trained on the Bank Credit Card dataset, following rigorous preprocessing steps including handling of missing values, outlier removal, normalization, and feature selection using Recursive Feature Elimination (RFE). The XGBoost classifier achieves a high accuracy of 99.69%, along with 99.33% precision and recall, an F1score of 98.61%, and an AUC value of 0.997. Comparative analysis with Neural Network, LSTM, Random Forest and Decision Tree models further validates XGBoost’s superiority. While XGBoost provides excellent predictive capability, its computational complexity may pose challenges in large-scale real-time deployments. The proposed model presents a robust, scalable approach for credit scoring in modern financial ecosystems.