Effective mentorship has a significant impact on professional and educational outcomes. However, finding optimal mentor-mentee pairs that maximize compatibility while satisfying multiple constraints presents a complex optimization problem. A comprehensive comparative analysis of four algorithmic approaches to mentor-mentee matching: Euclidean distance-based matching, K-means clustering, Genetic Algorithm (GA), and Deferred Acceptance (DA) is presented in this study. Utilizing a dataset of 8,000 potential mentors and mentees with 13 matching features, we evaluated algorithm performance across 100 randomized iterations under varying constraint conditions and one-sided matching directionality. Experimental results revealed distinctive performance profiles: Euclidean demonstrated the best balance of computational efficiency (average 2.13 s execution time) and match quality under moderate constraints; K-means showed initial promise but rapidly degraded with increasing constraints (failing after 7 constraints); GA achieved the highest similarity scores (91.78%) but with substantial computational overhead (648.50 s); and DA maintained strong validity (93.58%) with moderate efficiency. This study demonstrated that algorithm selection for mentor-mentee matching must prioritize system-specific requirements, computational resources, and quality priorities, with simpler algorithms often outperforming complex approaches under manageable constraint loads.