Scientist for Machine Learning
Research, Earth System Modelling, Innovation Platform
Summary:
Sara Hahner is a machine learning scientist working on representing the earth system components in the data-driven weather forecasting system at ECMWF.
She has a scientific background in applied mathematics and has studied machine learning applications in other domains, for example, the representation learning for 3D surface meshes and the postprocessing of car crash simulations.
Professional interests:
- Machine Learning for Earth System Modelling
- Representing different components in an Earth system model
- Data-driven ocean, sea ice, and wave modelling
Career background:
- since 2024: Scientist for Machine Learning at ECMWF
- Ph.D (Dr. rer. nat.) Mathematics, University of Bonn, Germany (2024). Thesis on "Low-dimensional Representations for Diverse Collections of 3D Surface Meshes" (supervised by Jochen Garcke)
- 2020 - 2024: Machine Learning Scientist, Fraunhofer Scientific Computing and Algorithms Institute (SCAI) in Sankt Augustin, Germany
- 2022: Research Internship, École Polytechnique, France. Visiting research group of Maks Ovsjanikov
- M.Sc. Computer Science, University of Bonn, Germany (2019). Major: Intelligent Systems, Thesis on "Analyzing and Predicting Simulation Results using Oriented Bounding Boxes and Recurrent Neural Networks"
- B.Sc. Mathematics, University of Bonn, Germany (2017). Major: Numerical Analysis, Minor: Computer Science, Thesis on "Analysis of an Inverse Mapping of Nonlinear Dimensionality Reductions with Application on Simulation Data"
- 2025
- Lorenzo Zampieri, Harrison Cook, Rachel Furner, Sara Hahner, Florian Pinault, Baudouin Raoult, Nina Raoult, Mario Santa Cruz, Matthew Chantry (March 2025) Coupling approaches for data-driven Earth system models. DOI: 10.5194/egusphere-egu25-4499
- Rachel Furner, Rilwan Adewoyin, Mario Santa Cruz, Sara Hahner, Sarah Keeley, Kristian Mogensen, Lorenzo Zampieri (March 2025) Developing a data-driven global ocean model at ECMWF . DOI: 10.5194/egusphere-egu25-11883
- Sara Hahner, Jean Bidlot, Josh Kousal, Lorenzo Zampieri, Matthew Chantry (March 2025) Representing waves in ECMWF’s data-based forecasting system AIFS. DOI: 10.5194/egusphere-egu25-11946
- Sara Hahner (March 2025) The AIFS: ECMWF’s data-driven weather forecasting system. DOI: 10.5194/egusphere-egu25-12077
- 2024
- Sara Vera Hahner, Jochen Garcke, Martin Rumpf (May 2024) Low-dimensional Representations for Diverse Collections of 3D Surface Meshes.
- Hahner S., Attaiki S., Garcke J., Ovsjanikov M. (May 2024) Unsupervised Representation Learning for Diverse Deformable Shape Collections, Proceedings - 2024 International Conference on 3D Vision, 3DV 2024. DOI: 10.1109/3DV62453.2024.00158
- 2022
- Hahner S., Kerkhoff F., Garcke J. (May 2022) Transfer Learning using Spectral Convolutional Autoencoders on Semi-Regular Surface Meshes, Proceedings of Machine Learning Research.
- Hahner S., Garcke J. (May 2022) Mesh Convolutional Autoencoder for Semi-Regular Meshes of Different Sizes, Proceedings - 2022 IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2022. DOI: 10.1109/WACV51458.2022.00240
- 2020
- Hahner S., Iza-Teran R., Garcke J. (May 2020) Analysis and Prediction of Deforming 3D Shapes Using Oriented Bounding Boxes and LSTM Autoencoders, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). DOI: 10.1007/978-3-030-61609-0_23