The Certified Independent Study and Internship Program (MSIB) is part of the Merdeka Belajar Kampus Merdeka (MBKM) initiative, allowing students to convert up to 20 credits from their activities. However, this process is often challenging as universities must match the MSIB syllabus with the curriculum syllabus of the Informatics Engineering program at ITN Malang. Differences in terminology and specific rules between the two syllabi make the process time-consuming and confusing. This study aims to develop a recommendation system to simplify the course conversion process. The system employs a Natural Language Processing (NLP) model based on Transformers to capture textual context and a cosine similarity algorithm to measure the similarity between syllabi. The system evaluation using classification metrics achieved an accuracy of 67%. The precision score reached 71% for the majority class and 50% for the minority class, while recall was 83% and 33%, respectively. The weighted average produced an f1-score of 0.65, indicating satisfactory performance despite class imbalance. These results demonstrate the system's potential to provide reliable recommendations, although further optimization is required to improve performance for the minority class.