We are seeking you to support our research on AI-based spatial reconstruction of urban green roof soil moisture. Your primary focus will be to transfer observed moisture dynamics from instrumented reference roofs to nearby, unsensed roofs using generative AI methods and complementary spatiotemporal interpolation techniques. The work is connected with the Helmholtz Water Security Initiative and contributes to the Helmholtz Solution Lab BlueGreen L.E.

The overarching goal is to develop robust generative models that reconstruct soil moisture fields across networks of green roofs, accounting for various effects, such as microclimate, roof properties, or hydrological processes. The models will fuse multi-modal data (in-situ sensors, meteorological observations, urban morphology and roof metadata, and available remote sensing) to produce consistent, uncertainty-aware estimates suitable for operational water-sensitive urban design and stormwater management.

Project work will be conducted in close cooperation with Solution Lab partners, including municipal stakeholders, ensuring transferability of methods across different urban settings and alignment with real-world decision needs in Blue-Green infrastructure planning.

Organizational unit

Institut für Meteorologie und Klimaforschung Atmosphärische Umweltforschung (IMKIFU)

Job description

Your main tasks will be to:

• conceptualize and set up the spatial reconstruction framework based on generative AI
• develop and benchmark AI-based methods (GAN, diffusion, transformer) and standard interpolation techniques (graph/spatiotemporal interpolation, kriging/GPs)
• integrate heterogeneous data sources and coordinate joint model development with project partners
• assist in the setup and maintenance of the roof sensors including data curation
• summarize, present, and publish the project results

A part-time position is possible.

Starting date

01.04.2026

Specifically, you have:

• a PhD degree in the field of Artificial Intelligence or the wider field of Machine Learning, ideally in the context of hydrology, meteorology or associated fields
• strong skills in developing and training deep neural networks, ideally using pytorch;
• expertise with handling large datasets and machine learning workflows;
• basic knowledge of hydrology, meteorology and remote sensing;
• demonstrated history of publications in peer-reviewed journals;
• willingness to travel and to assist in the installation and maintenance of soil moisture sensors.

You will also need to have a good command of written and spoken English.