Personalized course selection has become increasingly important in higher education as students are presented with a growing range of courses and specialization options. Conventional recommendation approaches generally rely on academic performance, previous course choices, or stated preferences, which may not adequately capture the cognitive abilities and subject specific competencies required for successful learning. This paper proposes a competency aware machine learning framework for personalized course recommendation that integrates student subject knowledge, cognitive skills, and expert defined course requirements. The proposed framework first constructs an individual student competency profile by combining subject level knowledge assessment with cognitive skill assessment covering dimensions such as logical reasoning, analytical thinking, problem solving, pattern recognition, and computational thinking. In parallel, domain experts define the relative importance of these skills for different courses through a course skill knowledge model. The resulting student and course representations are used to determine competency–course alignment and generate personalized recommendation scores. Machine learning models are then employed to learn the relationship between student characteristics and suitable course choices, while an explainable recommendation component identifies the skills contributing to each recommendation. The framework is evaluated against progressively enriched baseline models using both predictive and recommendation-oriented measures, including accuracy, precision, recall, and F1 score, Precision@K, Recall@K, NDCG@K, and MRR. An ablation analysis is further used to examine the con attribution of cognitive skills, subject competency, and expert knowledge. The proposed approach aims to provide recommendations that are not only personalized but also competency aligned and interpretable, thereby supporting more informed course selection in computer science education.