This study proposes an AI-driven assessment framework for evaluating students’ academic competence, addressing the limitations of traditional GPA-based methods, which lack personalization and comprehensiveness. The framework integrates Educational Data Mining (EDM), machine learning, and explainable AI (XAI) to enhance transparency and rigor. It starts with K-Means clustering to automate student performance labeling, reducing subjectivity. Five machine learning models—XGBoost, SVM, Random Forest, K-Nearest Neighbor, and Logistic Regression—are trained, with the optimal SVM-based model selected through evaluation. To ensure interpretability, the Dual-Focus Explainer Framework (DFEF) combines SHAP and LIME for both global feature importance and individual decision insights, promoting transparency in educational decision-making. Experimental results show high alignment with expert evaluations, validating the framework’s effectiveness and improving the reliability and acceptability of academic assessments.