Job description
The Scientific Computing Center is the Information Technology Center of KIT.
Earth system models (ESMs) are highly complex software systems that have often developed over several decades. In addition to the core model that computes meteorological processes, an ESM includes a wide range of specialized submodels - such as those for cloud microphysics or atmospheric chemistry - interconnected through well-defined interfaces. The computational demands of ESMs are immense, but their high degree of parallelism enables long-term climate simulations on modern HPC clusters. Continuously adapting these models to the latest hardware and software developments remains an ongoing challenge.
Your responsibilities in this area will include the following tasks (in close collaboration with domain scientists):
1. Ensuring the sustainability of community ESM codes, and enabling the adoption of modern HPC systems
2. Enabling and improving high-performance computing capabilities of ESM codes, including performance optimization
3. Contributing to porting ESM codes to heterogeneous HPC architectures, including GPU-accelerated systems
4. Participation in model developments with regard to the further development of an improved simulation of the Earth system
5. Participate in the NHR ESM community through software development support and user outreach
6. Carrying out own research in atmospheric modeling, including running and analyzing ESM simulations
7. Participating in scientific conferences and contributing to peer-reviewed publications
Starting date
01.03.2026
Personal qualification
8. Degree (Master's level) and doctorate in atmospheric research, computer science, physics, engineering or a related field
9. Strong knowledge in the programming languages Fortran, C++, and Python
10. Solid background in build engineering, including Makefile development, compiler configurations, built optimization, and experience with Linux-based HPC clusters
11. Good software development practices, including version control with Git, documentation, and CI/CD workflows
12. Substantial experience with distributed- and shared-memory parallelization using MPI and OpenMP. Additional background in developing for heterogeneous architectures (e.g., CUDA or ROCm) is advantageous
13. Familiarity with modern deep learning frameworks (e.g., PyTorch or TensorFlow) and contemporary AI models, such as transformer-based models or convolutional neural networks, is advantageous
14. Strong communication and presentation skills, and the ability and motivation to work effectively in a team
15. Good written and spoken English skills; German language skills are an advantage
Curious about an exciting and versatile role in an agile team? Discover more about SCC as your professional place to be: