CV

Education

• Ph.D in Physics, University of Bremen, 2026
  • Supervisor: Prof. Veronika Eyring
  • Thesis: Interpreting and improving the radiation parameterization for ICON with machine learning
• M.Sc. in Physics, RWTH Aachen University, 2021
  • Supervisor: Prof. Martin Erdmann
  • Thesis: Autoencoder-extended Conditional Invertible Neural Networks for Unfolding of Radio Signals at the Pierre Auger Observatory
• B.Sc. in Physics, RWTH Aachen University, 2019
  • Supervisor: Prof. Martin Erdmann
  • Thesis: Refinement of Simulated Air Shower Detector Traces using Cycle-Consistent Adversarial Networks

Work experience

• Since March 2026: Postdoctoral Researcher
  • University of Lausanne
• Feb 2022 - Feb 2026: Research Assistant
  • University of Bremen

Skills

• Machine Learning
  • Supervised Learning
  • Generative Modelling
  • explainable AI
• Climate Modelling
  • Atmosphere
  • Radiation
  • ICON
• Programming
  • Python
  • LaTeX
  • Bash
  • Fortran
  • TypeScript/JavaScript
• Languages
  • German (native)
  • English (fluent)
  • French (basic)

Publications

Interpretable Machine Learning‑Based Radiation Emulation for ICON

Hafner, K., and Iglesias‑Suarez, Fernando and Shamekh, Sara and Gentine, Pierre and Giorgetta, Marco A. and Pincus, Robert and Eyring, Veronika (2023). "Interpretable Machine Learning‑Based Radiation Emulation for ICON." Journal of Geophysical Research: Machine Learning and Computation.e2024JH000501. DOI: 10.1029/2024JH000501

Autoencoder-extended Conditional Invertible Neural Networks for Unfolding Signal Traces

Erdmann, M., Hafner, K., Schulte, J. & Straub, M. (2023). "Autoencoder-extended Conditional Invertible Neural Networks for Unfolding Signal Traces." Journal of Physics: Conference Series. DOI: 10.1088/1742-6596/2438/1/012072

Talks

Interpretable Machine Learning‑based Radiation Emulation for ICON

December 12, 2024

Talk at AGU Annual Meeting, Washington, D.C., USA

Improvement of the radiation parameterization of ICON by including subgrid‑scale cloud properties using machine learning

December 13, 2023

Talk at AGU Annual Meeting, San Francisco, CA, USA

Autoencoder‑extended Conditional Invertible Neural Networks for Unfolding Signal Traces

November 29, 2021

Poster at 20th International Workshop on Advanced Computing and Analysis Techniques in Physics Research, online/ Daejeon, South Korea

Teaching

Teaching experience 2

Teaching experience 1

Service and leadership

• Reviewer for JAMES
• Convener for Machine Learning for Climate Science at EGU General Assembly 2026