The major contributor to global mortality is cardiovascular disease, posing a formidable challenge to the global healthcare system. Heart disease often develops and progresses without noticeable symptoms, emphasizing the need for earlier detection to prevent severe outcomes. AI models provide tools for accurate diagnosis, data analysis, and predictive diagnosis of people who may be suffering from heart disease. But increasingly, solutions like Explainable AI (XAI) are gaining traction. XAI demystifies the decisions of AI models, making them transparent and understandable to the point where they boost confidence in the context of AI diagnostics by healthcare professionals; the AI then provides support to professionals to decide the best course of action for each patient based on the analysis. XAI techniques such as SHAP, LIME, QLattice and Anchor are popular, and have been used in this study. We analysed a heart disease dataset collected from a multispecialty hospital in India, which is publicly available on Mendeley. We used ML models, including RF, Logistic Regression, KNN, XGBoost, DT and SVM. These models were optimized and stack-ensembled to form the ‘CARDIACX’ model to predict the risk of heart disease. The models were fine-tuned using Grid Search, Random Search, and Bayesian Optimization methods, achieving promising results with Random Forest achieving an AUC = 0.99 and accuracy of 98.5%, outperforming all the other models and demonstrating robust performance on various metrics such as AUC, PR Curve, Log Loss, Jaccard Score, and MCC. This study provides a reliable and transparent framework for early detection of heart disease, providing useful information and enabling patients to receive timely and personalized care.