My doctoral research focuses on the application of artificial intelligence and explainable machine learning methods to oceanographic processes, with the goal of advancing scientific understanding and improving predictive capabilities.

Research Activity

I applied Reservoir Computing to investigate whether El Niño variability arises from a self-sustained oscillatory mechanism or is driven by stochastic forcing. In a complementary study, I examined why Reservoir Computing achieves superior forecasting performance compared to classical dynamical models. This work resulted in two first-author publications. I also contributed to collaborative studies investigating the mechanisms underlying Central Pacific and Eastern Pacific El Niño events.

I worked on the application of convolutional neural networks (CNNs) to reconstruct the Atlantic Meridional Overturning Circulation (AMOC) strength from observational data, in collaboration with Dr. Simon Michel, Valérian Jacques-Dumas, Dr. René van Westen, and Prof. Dr. ir. Henk A. Dijkstra, resulting in a co-authored publication. In a separate project, I developed a CNN-based framework to estimate the distance of the AMOC to a potential collapse.

Teaching and Mentorship

I served as a Teaching Assistant for the Advanced Topics in Climate Dynamics course (Machine Learning for Climate module), where I contributed to designing tutorials, assignments, and projects focused on applying machine learning and deep learning techniques to climate science problems, and supported students in implementing these methods.

In addition, I was a Teaching Assistant for the Earth System Modeling course, assisting students in understanding climate models and designing numerical experiments.

I co-supervised two Master’s theses:

• one on the use of Neural ODEs to improve AMOC representation in intermediate-complexity models
• one on the application of 3D CNNs for AMOC strength forecasting