PhD Fellow: Data Science Environment & Life Sciences

The Helmholtz School for Integrated Data Science in Environmental and Life Sciences (IDEAS) connects the domain-science expertise of UFZ and HZDR with the data/information science strength of Leipzig University (LU) and TU Dresden (TUD), supported by CASUS as an interdisciplinary bridge. IDEAS is part of the Helmholtz Information & Data Science Schools under the Helmholtz Data Science Academy (HIDA).

IDEAS advances and applies modern data science (e.g., machine learning, explainable AI, uncertainty quantification, and AI-ready FAIR data and research data management) to complex challenges in environmental and life sciences.

IDEAS offers structured, interdisciplinary supervision and training, including joint supervision across disciplines, a Thesis Advisory Committee (TAC), a tailored curriculum, and cohort activities (seminars, hackathons, retreats), plus strong career development and networking through the IDEAS/HIDA ecosystem.

This collective call includes 5 PhD topics, of which 4 positions will be funded. Applicants can be considered for multiple projects and will be matched through a structured selection and ranking process.

• Flood Lens

Severe storms and floods cause large damages, and when occurring simultaneously over different regions, emergency responses and relief might be additionally strained. This project will combine advanced deep learning architectures and causal representation learning frameworks coupled with explainable AI to develop physically interpretable, robust, and trustworthy data-driven seasonal and sub-seasonal forecasts of spatially co-occurring flood events and their large-scale atmospheric precursors.

• SoilCloudAI

Soil moisture can influence clouds, droughts, and heatwaves, but these feedbacks remain difficult to quantify because they are nonlinear, spatially connected, and strongly regime dependent. This project will develop interpretable and probabilistic graph-based AI methods to identify soil moisture-cloud feedback pathways from in-situ, satellite, reanalysis, and climate-model data. By combining Earth system science with modern data science, it aims to improve our understanding of land-atmosphere feedbacks and develop transferable methods for complex spatiotemporal environmental datasets.

• TRACE-GBM

Can artificial intelligence design the next generation of radiotracers for brain tumors? Glioblastoma remains one of the deadliest human cancers and urgently requires improved tools for molecular imaging and targeted therapy. This project combines state-of-the-art generative protein design, machine learning, radiochemistry, and PET imaging to develop novel mini-protein binders against glioblastoma biomarkers. Through an iterative design-build-test-learn framework, computationally designed binders will be experimentally validated and translated into radiopharmaceutical probes for molecular imaging and theranostic applications.

The project aims to establish a new paradigm for data-driven radiotheranostic development at the interface of AI, protein engineering, and neuro-oncology.

• SafeBEEP

The loss of pollinating insects is caused by the utilization of plant protection productions with unintended side effects. Using data science and ai we intend to predict the elimination of plant protection products by the microbiome of pollinators, and so we could keep the bees safe. The prediction of transformation is already established for other microbiomes (take a look) and now we want to combine it with graph representation of chemical reactions.

• Digit Health

Continuous metabolic sensing technologies now enable high-frequency monitoring of metabolites such as lactate, generating rich physiological time-series data that capture tissue metabolism and adaptation beyond the capabilities of conventional sparse sampling. This PhD project aims to develop novel digital markers of tissue health from continuous metabolic sensing data by combining advanced biosensing technologies with machine learning and data science. The research follows a structured workflow:
Generate Data → Expand Biological Measurements → Learn Digital Markers → Predict…