The unavailability of Waste Sort Recyclables (WSR) is a serious waste management problem in Rivers State, Nigeria. WSR is a deep learning process that involves the classification of waste into four recycling categories which include glass, paper, metal and plastic. Secondly, there is need for a recommender system for relevant waste management agencies in Nigeria. In this study, we developed an improved model for Waste Management Recommender System (IWMRS) in Nigeria using Deep Learning approach. Software Development and Lifecycle Methodology (SDLC) was utilized in this approach. Furthermore, we implemented with Hypertext Pre-processor, JavaScript Programming Languages and MySQL Relational Database as backend. The parameters for our results performance achieved an overall performance rate of 94% when compared with the most recent Waste Management System. The parameters for the comparative analysis included Time Complexity (TC), Life-Cycle Assessment (LCA), Benchmarking (B), Multi-Criteria Decision Making (MCDM), Risk Assessment (RA), Cost Benefit Analysis (CBA) and Speed (S) which was also presented as TC, LCA, B, MCDM, RA, CBA, S= 20, 36, 14, 10, 7, 2, 5 respectively as compared with the existing parameters values of 14, 31, 14, 10, 7, 2 and 5 and further confirmed outperformance of the existing system by the proposed system. The obtained results show also the importance of Deep Learning techniques in Recommender Systems. This is because we live in a World of Information and Big Data. In addition, we boldly recommend this study to seekers of waste management information through recommender systems that utilizes Deep Learning Techniques.