<![CDATA[Teaching ratios at the junior high school level often poses a challenge for students who need a deep conceptual understanding to distinguish between direct and inverse proportions through real world contexts. Therefore, this study aims to produce a valid, practical, and effective Deep Learning based instructional module on ratios. The research method employed a Research and Development (R&D) approach using the 4D development model, which includes the Define, Design, Develop, and Disseminate stages, involving 25 students from MTs Hasyimiyah Bungah Gresik. The instruments used included an expert validation sheet, a teacher response questionnaire, a student response questionnaire, and a summative assessment. Based on the research results, the developed teaching module meets the criteria for validity, practicality, and effectiveness. The validation test results met the validity criteria, with an average score of 3.94 from two validators, falling into the highly valid category. Furthermore, the practicality test results met the practicality criteria, with an average score of 3,92 from the teacher response questionnaire and an average score of 3,90 from the student response questionnaire thus, the combined average score of 3,91 from both questionnaires falls into the highly practical category. Meanwhile, regarding the effectiveness test results, the module met the effectiveness criteria with an average score of 83,2 on the summative assessment test, placing it in the highly effective category. Therefore, the teaching module using the Deep Learning approach can be implemented and is recommended to teachers to stimulate conceptual understanding and enhance student engagement through more meaningful learning in the instructional process. The novelty of this study lies in the development of a ratio module that integrates Deep Learning (Mindful, Meaningful, and Joyful) in accordance with the Merdeka Curriculum, which transforms procedural learning into a deep mastery of concepts. Additionally, previous research has not extensively explored the Deep Learning approach to ratio material.]]>
