Integration Deep Learning and Tarbiyah Through Multidimensional Projects At SMPIT Al-Izzah Kotabaru

Abstract

The growing demands of twenty-first-century education require Islamic schools to integrate advanced cognitive competencies with moral and spiritual formation in ways that are pedagogically coherent and contextually meaningful. This study examines the implementation of the Multidimensional Project Model at SMPIT Al-Izzah Kotabaru as an innovative framework for merging Deep Learning principles with the Islamic educational paradigm of tarbiyah. Employing a descriptive qualitative case study design, data were collected through classroom observations, semi-structured interviews with Grade 8 teachers and students, and documentation analysis, supported by triangulation to ensure credibility. The findings indicate that the model effectively bridges disciplinary boundaries by positioning scientific inquiry within spiritually grounded learning experiences. Students demonstrated noticeable improvement in collaboration, task distribution, and conflict resolution key components of 21st-century skills. Importantly, these improvements were closely associated with the internalization of Islamic values such as amanah (trustworthiness), ihsan (excellence), and khalifah fil ard (stewardship of the Earth). Through ecosystem-based projects, students perceived environmental responsibility not only as a scientific requirement but also as an expression of religious duty, thereby strengthening their sense of purpose and accountability. The study concludes that integrating Deep Learning with tarbiyah-oriented project work produces holistic learning outcomes that cultivate both intellectual depth and spiritual consciousness. The Multidimensional Project Model serves as a replicable micro-level innovation for Islamic schools implementing the Merdeka Curriculum, offering a practical strategy to unify academic achievement and character development. Further research is recommended to develop measurable indicators for spiritual competencies and to explore cross-grade expansion of the model.