Accurately predicting student performance has become a crucial task in higher education, as it enables early identification of at- risk learners and supports datadriven academic interventions. This study proposes a predictive framework based on Multi-Layer Artificial Neural Networks (ANNs) to model the complex relationships between student attributes and academic outcomes. Input features such as demographic information, academic history, attendance, learning behavior, and socio-economic background are used to train the model. The proposed ANN architecture leverages multiple hidden layers to capture nonlinear dependencies and improve prediction accuracy compared to conventional machine learning methods. Experimental results demonstrate the effectiveness of the model in predicting student grades and classifying performance levels. This research highlights the potential of deep learning approaches in enhancing student support systems, personalized learning strategies, and academic advising, ultimately contributing to improved educational quality and institutional decision-making.