This study was carried out to advance the predictive capabilities for entrepreneurial competency among university students. Building upon prior models including K-nearest neighbors, random forest, and logistic regression, which fell short of the 70% accuracy threshold, we introduced XGBoost and LightGBM gradient boosting algorithms. Both surpassed the established benchmarks with XGBoost achieving a remarkable 79% accuracy and outperforming LightGBM's 71%. This improvement underscored LightGBM's prowess in handling voluminous datasets. We proposed a transformative blueprint for universities to enhance curricula, cultivating entrepreneurial ecosystems. By exploring educational sectors and non-entrepreneurial scores, institutions can create environments conducive to entrepreneurial endeavors and address gender disparities. Unraveling reasons behind reluctance to pursue entrepreneurship empowers decision-makers to refine curricula and foster a more supportive environment for aspiring entrepreneurs.