Employee turnover is a persistent challenge for HR departments, especially within large organizations that utilize complex enterprise systems like SAP SuccessFactors and SAP HCM. This study presents a machine learning-based predictive framework to identify potential employee exits before they occur, leveraging historical HR data spanning performance, compensation, demographic, and behavioral metrics. By training and validating various machine learning models—including Random Forest, Gradient Boosting, and Neural Networks—on anonymized employee datasets extracted from SAP modules, we aim to uncover patterns and leading indicators of voluntary and involuntary turnover. The methodology incorporates data preprocessing, feature selection, class balancing, and model interpretability strategies such as SHAP values. Our results demonstrate that the Random Forest model achieved the highest accuracy at 86%, with critical predictors being low engagement scores, lack of internal mobility, and stagnant compensation growth. The study concludes by offering a framework for proactive retention strategies and outlines implications for integrating AI-driven insights directly into HR workflows. These findings contribute to the evolving practice of predictive HR analytics and establish a replicable pipeline for real-time turnover forecasting using enterprise resource data.