With the advent of blended learning models, there is more demand for custom made learning programs. This paper puts forward a learning path algorithm based on deep learning, which can provide customized and real-time adjustable learning path for learners. The powerful capabilities of deep learning frameworks allow this algorithm to intelligently craft the optimal learning experience according to every learner’s knowledge reserves, learning processes, and behavioral habits, thus maximising learning efficiency and engagement. The study includes areas of focus, including the design of algorithms, processing of data, path generation, and research methods, and uses real datasets to confirm the accuracy and efficacy of your algorithm. It confirms that the algorithm plays an important role in optimizing learning experience and improving learning effect, which lays a solid foundation for the intelligent upgrading of hybrid education platforms.