Helping students design a structured learning path plays an important role in higher education, especially when semester-wise recommendations align with student academic goals, career aspirations, and graduation requirements. Indeed, an effective recommendation will help students improve their learning results by promoting personalization and providing informed decisions based on data-driven insights such as students’ performance in history, learning preferences, and impactful factors. This is especially necessary for students with poor academic performance so that they can be provided with a personalized academic pathway to escape this situation. Therefore, in this paper, we propose a grade prediction model and then recommend personalized academic pathways at the program level. The solution is based on grade prediction using deep learning models to recommend the pathways with courses suitable for students to study in the remaining semesters for better performance and a higher chance of timely graduation. The experimental results on the benchmark dataset show that suggesting academic pathways based on our predicted grades brings more positive results to students compared to other approaches.