Florian Scheidl

I studied maths and data science at ETH Zurich, completing my master's studies in Torsten Hoefler's group researching higher-order graph neural networks. After my studies, I investigated neural compression of climate data combined with traditional compression techniques involving wavelet-transforms. Recently, I worked on energy forecasting at Danish energy tech startups, taking on MLOps, data engineering and data science roles. Now I am excited to work on HPC capabilities for the WeatherGenerator project.

Huang, Langwen, Luigi Fusco, Florian Scheidl, et al. “Error Bounded Compression for Weather and Climate Applications.” arXiv Preprint arXiv:2510.22265, 2025.

Besta, Maciej, Florian Scheidl, Lukas Gianinazzi, et al. “Demystifying Higher-Order Graph Neural Networks.” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025.

Gianinazzi, Lukas, Alexandros Nikolaos Ziogas, Langwen Huang, et al. “Arrow Matrix Decomposition: A Novel Approach for Communication-Efficient Sparse Matrix Multiplication.” 2024, 404–16.

Besta, Maciej, Pawel Renc, Robert Gerstenberger, et al. “High-Performance and Programmable Attentional Graph Neural Networks with Global Tensor Formulations.” 2023, 1–16.

Besta, Maciej, Patrick Iff, Florian Scheidl, et al. “Neural Graph Databases.” 2022, 31–1.

Fritschi, Lea, Johanna Hedlund Lindmar, Florian Scheidl, and Kerstin Lenk. “Neuronal and Astrocytic Regulations in Schizophrenia: A Computational Modelling Study.” Frontiers in Cellular Neuroscience 15 (2021): 718459.