Recommendation system for connecting students with relevant alumni can greatly enhance mentorship and career guidance. Our approach leverages advanced natural language processing (NLP) techniques to go beyond traditional keyword matching, which often misses nuanced skill relationships. By using a Sentence Transformer model, we transform both students’ and alumni’s skills into high-dimensional vectors that capture semantic similarities between their expertise. The system calculates cosine similarity between these embeddings, measuring how closely aligned the skill sets are. This similarity score forms the foundation of a graph-based model, where students and alumni are represented as nodes, and edges are weighted by the strength of their skill match. To ensure relevance, we only create connections between individuals within the same department, allowing for more targeted recommendations. By personalizing alumni suggestions based on skill alignment rather than simple keyword matching, this system helps students find mentors who can provide meaningful, relevant career advice, fostering better professional relationships and opportunities.