<![CDATA[This study analyzes the implementation of academic supervision through a Deep Learning approach in enhancing the pedagogical competence of Islamic Religious Education (PAI) teachers at SDN 224 Palembang. The background of this research is the gap between the demands of 21st-century pedagogical competence and the reality of supervision practices that remain conventional and administrative in elementary schools. The study aims to describe PAI teachers' pedagogical competence, analyze the Deep Learning approach in supervision, and examine its implementation to enhance pedagogical competence. Using a descriptive qualitative approach with a case study design, data were collected through non-participant observation, in-depth interviews, and documentation analysis. The findings reveal three main themes. First, PAI teachers' pedagogical competence is still dominated by conventional methods with minimal technology integration, while academic supervision focuses on administrative compliance without reflective feedback. Second, the Deep Learning approach is understood as a framework for analyzing learning depth including question quality, cognitive interaction, and value contextualization yet remains unimplemented in supervision instruments. Third, Deep Learning-based supervision has the potential to enhance pedagogical competence through data-driven feedback, but is constrained by technological infrastructure, teachers' digital literacy, administrative burden, and the absence of a pedagogical collaboration culture. This study implies the necessity of transforming the supervision paradigm from administrative to reflective-analytical, strengthening technology-based instructional leadership, and developing supervision instruments based on Deep Learning indicators. Abstrak Penelitian ini menganalisis implementasi supervisi akademik melalui pendekatan deep learning dalam meningkatkan kompetensi pedagogik guru Pendidikan Agama Islam (PAI) di SDN 224 Palembang. Latar belakang penelitian ini adalah kesenjangan antara tuntutan kompetensi pedagogik abad 21 dengan realitas praktik supervisi yang masih bersifat konvensional dan administratif di sekolah dasar. Penelitian bertujuan mendeskripsikan kompetensi pedagogik guru PAI, menganalisis pendekatan deep learning dalam supervisi, serta mengkaji implementasinya untuk peningkatan kompetensi pedagogik. Menggunakan pendekatan kualitatif deskriptif dengan studi kasus, data dikumpulkan melalui observasi non-partisipan, wawancara mendalam, dan analisis dokumentasi. Hasil penelitian mengungkap tiga temuan utama. Pertama, kompetensi pedagogik guru PAI masih didominasi metode konvensional dengan integrasi teknologi minimal, sementara supervisi akademik terfokus pada pemeriksaan administratif tanpa umpan balik reflektif. Kedua, pendekatan deep learning dipahami sebagai kerangka analisis kedalaman pembelajaran meliputi kualitas pertanyaan, interaksi kognitif, dan kontekstualisasi nilai namun belum terimplementasi dalam instrumen supervisi. Ketiga, implementasi supervisi berbasis deep learning berpotensi meningkatkan kompetensi pedagogik melalui umpan balik berbasis data, namun terkendala infrastruktur teknologi, literasi digital guru, beban administratif, dan minimnya budaya kolaborasi pedagogis. Penelitian ini mengimplikasikan perlunya transformasi paradigma supervisi dari administratif menuju reflektif-analitis, penguatan kepemimpinan pembelajaran berbasis teknologi, serta pengembangan instrumen supervisi berbasis indikator deep learning.]]>
