This study aims to evaluate the management of deep learning in junior high schools in Ambon City using the Context, Input, Process, and Product (CIPP) evaluation model. The study employed an evaluative qualitative approach to obtain an in-depth description of how deep learning is planned, organized, implemented, assessed, and followed up at the school level. Data were collected through semi-structured interviews, classroom observation, and documentation involving principals, teachers, and school management staff from three public junior high schools in Ambon City. The data were analyzed using the interactive model of data reduction, data display, and conclusion drawing, while credibility was strengthened through source triangulation, technique triangulation, and member checking. The findings show that, in the context component, school visions and missions are generally aligned with deep learning principles, although the translation of these principles into classroom practice remains uneven. In the input component, teacher qualifications, support staff, facilities, learning resources, and training programs are available but require further strengthening, particularly in authentic assessment, contextual material development, and technology integration. In the process component, teachers have begun to use project-based learning, problem-based learning, inquiry, reflection, portfolio assessment, and collaborative planning, but implementation is constrained by time, literacy levels, heterogeneous student abilities, and passive learning habits. In the product component, students show emerging improvements in critical thinking, independence, collaboration, and contextual problem solving. The study recommends continuous professional development, stronger instructional supervision, better facility optimization, and a participatory evaluation system for deep learning management.