I am a data scientist in André Rendeiro’s Spatial and Computational Biology group at CeMM. I work in computational pathology on H&E: stain variability, and GPU pipelines.

I care about making large linear algebra cheap on GPUs: unstructured sparse matrix multiply, and mixed-precision kernels that stay correct. The same instinct shows up in clustering backends and histopathology tooling—native CUDA instead of a slow fallback, and software that actually installs.

At the University of Vienna I lecture in the Research Group Theory and Applications of Algorithms, and my research is graph neural networks under resource constraints: mixed-precision quantization (MixQ, IEEE ICDE 2025), GNN loss landscapes (ICANN 2025), and redundancy in GNNs (ECML PKDD 2024). Courses include Numerical High Performance Algorithms, Software Tools for Computational and Data Science, and Mathematical Foundations of Computer Science.

Across both labs I care about making the arithmetic cheap: custom floating-point formats, operation counts, sparse and block-scaled GPU kernels. Before CeMM I was a university assistant at UniVie and an R&D engineer at Huawei.