My research focuses on machine learning and its applications in medical image analysis, with a particular interest in understanding the representations learned by deep neural networks and guiding their development. This led me to explore self-supervised learning for 3D medical imaging, where I aim to develop general-purpose representations applicable to various downstream tasks. Notably, I recently collaborated with the German start-up FLOY to develop a large pre-trained brain MRI model. I am also currently leading the development of pre-training methods for The Human Radiome Project, a pilot project of the Helmholtz Foundation Model Initiative, which seeks to create a vision foundation model for 3D radiology. Furthermore, I am currently working on curating a large-scale dataset, establishing a benchmark, and maintaining a comprehensive repository of the most relevant 3D SSL methods to accelerate the 3D medical SSL community.