Deep clustering has emerged as a powerful paradigm for uncovering latent structure in complex, high-dimensional data, yet its adoption in educational data analysis remains limited. This study explores the application of Deep Embedded Clustering (DEC) to the Moroccan TIMSS 2019 Grade 8 dataset, which combines student achievement with rich contextual information from students and schools. By jointly learning feature representations and cluster assignments, the proposed approach aims to identify meaningful student profiles that go beyond simple performance-based groupings. The resulting clusters reveal differentiated patterns in mathematics and science achievement, offering insights into student heterogeneity within a large-scale assessment context. Overall, the findings illustrate the potential of deep clustering methods to support exploratory educational analytics and contribute to a deeper understanding of learning profiles in complex educational datasets.