This session will explore how graph neural networks (GNNs) can help enhance recommender systems and personalisation algorithms for e-commerce and entertainment industries.
Mariia will talk about how GNNs were employed at Zalando (biggest European fashion e-commerce platform) to generate dynamic user and content embeddings, capturing complex relational data beyond traditional static features. By focusing on the relational context within user-content interactions, GNNs have improved the prediction of click-through rates (CTR), leading to more tailored and engaging user experiences. Attendees will gain insights into the architecture of the GNN model, the methodologies for training and integrating graph-based embeddings into existing systems, and the tangible benefits observed.
This session is ideal for professionals interested in cutting-edge recommender systems and the practical applications of GNNs in large-scale digital platforms.
Speaker
Mariia Bulycheva
Senior Machine Learning Engineer @Intapp, Previously Senior Applied Scientist @Zalando
Mariia Bulycheva is a Senior Machine Learning Engineer at Intapp, where she focuses on building large-scale knowledge graphs and developing intelligent systems powered by NLP - from classical approaches to cutting-edge large language models. Her work spans graph-based machine learning, building MCP servers and AI agents, bridging research with impactful real-world applications.
Before joining Intapp, Mariia was a Senior Applied Scientist at Zalando, where she designed and implemented a novel graph neural network architecture for content recommendation, introducing this methodology for the first time at the company. Her experience ranges from ranking multimodal content and optimizing long-term user engagement to building scalable AI systems for pricing and demand forecasting.
Slides are not available