Proceedings of the International scientific and practical conference ―Modern Science: Challenges and Perspectives‖ (February 9-11, 2026) / Publisher website: www.naukainfo.com. - London, United Kingdom, 2026. - 121 p.

45 INFORMATION TECHNOLOGIES AND SYSTEMS UDC 004.8:004.75:004.021 Vakhovskyi Oleksandr Independent Researcher Senior Software Engineer at SolidProfessor Mountain View, CA, USA APPLICATION OF RECOMMENDATION ALGORITHMS IN E- COMMERCE SYSTEMS: AN ENGINEERING APPROACH Abstract . This paper develops a classification of recommendation algorithms by data sources, mathematical models, and architectural characteristics. It analyses the operating principles of user-based and item-based collaborative filtering, matrix factorisation methods (SVD, ALS, NMF), content-based filtering and hybrid recommender models, as well as modern deep learning approaches. A comparative analysis of algorithms is conducted using criteria including recommendation accuracy, scalability, data requirements, interpretability, and the engineering complexity of integration. Key engineering challenges of deploying recommender systems are identified, including integration into distributed architectures, streaming data processing, ensuring low-latency inference, data privacy, and model lifecycle management within an MLOps infrastructure. Keywords: Commercial processes, collaborative filtering, deep neural network models, behavioural data, personalisation.