This study presents a practical and explainable hybrid deep–machine learning framework for predicting the Remaining Useful Life (RUL) of aircraft engines using the NASA C-MAPSS dataset. The proposed approach integrates a Long Short-Term Memory (LSTM) network to capture temporal degradation patterns with an Extreme Gradient Boosting (XGBoost) regressor for nonlinear RUL estimation. To enhance interpretability, SHapley Additive exPlanations (SHAP) were employed to identify the most influential features contributing to RUL predictions. The model was validated on all four C-MAPSS subsets (FD001–FD004), achieving RMSE scores ranging from 14.01 to 37.45, demonstrating its applicability across varying operating conditions and fault modes. The primary contribution is a framework that balances predictive accuracy with transparency. SHAP analysis, combined with gradient-based attribution, successfully linked latent model features to physical sensor groups (e.g., temperature, pressure, speed), enhancing trust and diagnostic insight. This work offers a practical framework for building more transparent and trustworthy prognostic models in industrial applications.