Purpose: The aim of the article is to develop a system for optimizing supply chain logistics processes using RFID technology. Design/Methodology/Approach: Machine learning algorithms such as Gradient Boosting, random forests, decision trees, RUS were used to solve the problem. An RFID reader and dedicated software were designed. Findings: The results of the conducted research show that the methods used, a procet application based on RFID technology, increases the company's efficiency in logistic processes. Practical Implications: The methods and system presented in the article can be used in the supply chain of logistics and manufacturing companies. Originality/Value: The novelty is the appropriate selection and use of machine learning methods for the designed system that optimizes the logistics processes of returns using RFID technology. A proprietary system consisting of a dedicated reader and IT application was designed.