Welcome

Hi, I’m Dr. Barrie Kersbergen, a Staff Scientist at Bol and researcher working on recommender systems and probabilistic forecasting.

Welcome to my academic website, here you’ll find my publications, talks, and open-source projects.

My PhD research expanded the boundaries of scalable session-based recommendation systems for large-scale e-commerce, addressing practical challenges like massive product catalogs, billions of interactions, sparse behavior data, and data quality issues. Conducted in close collaboration with Bol, a leading European e-commerce platform, this work bridges academic research and real-world deployment.

In my dissertation, Expanding Boundaries in Scalable Session-Based Recommendations, I demonstrate that nearest-neighbor methods can rival or even outperform deep learning models on large-scale datasets, especially under production constraints. I developed scalable algorithms for low-latency inference, robust benchmarking under resource limits, and efficient data debugging using Shapley values for data attribution.

This research delivered impactful solutions for large-scale recommender systems and laid the groundwork for more sustainable, efficient, and trustworthy recommendation technologies in practice.