Job description
We are seeking you to support our research on generative AI methods. You will mainly focus on the project 3D-GAIN (Realistic 3D Atmospheric Reconstruction for Generative AI Nowcasting of Precipitation and Irradiance using Remote Sensing and In-Situ Data). Additionally, your work will be embedded into the ongoing research on AI-based downscaling of rainfall fields in the newly established research group Hydrometeorological Sensing and Machine Learning at the KIT Campus Alpin.
The overarching goal of the project 3D-GAIN is to develop novel AI methods for generative 3D reconstruction of local atmospheric conditions to enhance the retrieval and nowcasting of rainfall and irradiance. This generative reconstruction will be based on multi-modal observations (satellite imagery, radar measurements, all-sky cameras, and in-situ sensors) and leverage a unified, physically consistent 3D representation of atmospheric conditions.
The project work in 3D-GAIN will be carried out in cooperation with the project partners at the DLR Institute of Solar Research and Energy Meteorology located in Almeria, Spain.
Your main tasks will be to:
1. Conceptualize and do initial setup of the 3D representation model
2. Develop an encoder-based data synchronization and a diffusion-based probabilistic sampling
3. Coordinate joint model development with the project partners
4. Summarize, present and publish the project results
Starting date
01.02.2026
Personal qualification
You have a PhD degree in the field of Artificial Intelligence or the wider field of Machine Learning, ideally in the context of meteorology or associated fields.
Specifically, you have:
5. strong skills in developing and training deep neural networks, ideally using pytorch;
6. expertise with handling large datasets and machine learning workflows;
7. basic knowledge of meteorology and remote sensing;
You will also need to have a good command of written and spoken English.