In Uganda, graduate unemployment persists due to a misalignment between university programmes and labour-market needs. This study introduces a Course Guidance System that predicts individual employment likelihood and informs academic choices. We assembled a graduate dataset of the university, Grade Point Average (GPA), course, A-level scores and real-time,community-sourced employment outcomes. After one-hot encoding and data cleaning, an XGBoost classifier was trained and evaluated via Leave-One-Out cross-validation. The model attained 98.9 % accuracy and a 99.1 %F1-score. Predicted probabilities were categorised as “Less Likely” (p < 0.50),“Likely” (0.50 ≤ p ≤ 0.75), and “Very Likely” (p > 0.75) to generate actionable guidance. Results highlight course of study, institution, and GPA as key employability factors. By leveraging up-to-date, community-collected data, the system overcomes static advisory limitations, offering dynamic insights for students, universities, and policymakers. Future work will integrate soft-skill metrics and employ explainable AI to enhance transparency.