My research evolves around geometric deep learning and the neural tangent kernel.
I work on mathematical aspects of AI, with focus on geometric deep learning. I also work on automorphic forms and their Fourier coefficients, with motivation from physics.
I work on field theory methods applied to modelling of deep neural networks.
I work on spherical computer vision and the theory of equivariant neural networks.
I am interested in the synergy between equivariant neural networks and data augmentation to build more sample-efficient and robust models.
My current research interests focus on equivariant neural networks, particularly on the trade-off between achieving equivariance through data augmentation and through manifestly equivariant architectures.
My research currently focuses on equivariant neural networks and the neural tangent kernel in the regime of large layers.
I am interested in mathematical aspects of local and global symmetries in neural networks, currently with a focus on transformers and their equivariant counterparts.
My main research interests are geometric deep learning and uncertainty quantification.
I work on making cars safer via autonomous driving.
