Fully funded PhD – Generative Foundation Models for ICU data (ERC Starting Grant, Innsbruck, Austria)

[Patrick Rockenschaub]

Hi all, 

We’re recruiting a fully funded PhD student to work on generative foundation models for ICU data in the ERC Starting Grant project GPT-MEDIC at the Medical University of Innsbruck (Austria). Women are particularly encouraged to apply.

Project
ICU data are irregular, continuous, and shaped by clinical decision-making. These characteristics makes cross-hospital generalisation hard. This PhD will develop generative pre-training methods for clinical event streams, aiming for models that transfer across settings and support multiple downstream prediction tasks.

Data & collaboration
You’ll work with a large multicentre ICU resource (~1M patients; ~33B events; ~half available from day one) and collaborate with partners at Amsterdam UMC, UCL, and University of Cambridge.

What we’re looking for

• MSc in CS/stats/ML (or a related quantitative field)
• Solid Python and experience with at least one deep learning framework (PyTorch/TensorFlow/JAX)
• Evidence you can complete technical work (thesis, paper, or GitHub/projects)

Clinical data experience is a plus but not required.

What we offer

• Fully funded position (3–4 years), start flexible
• Salary: €39k gross/year (30h/week; per university agreement)
• Access to modern GPU compute
• Conference travel + active publication support and close mentorship

Working language: English (German not required); relocation to Innsbruck expected

Apply (deadline: Sun, 25 Jan 2026)
Details: https://euraxess.ec.europa.eu/jobs/397624

Happy to answer any question via DM: https://www.linkedin.com/in/prockenschaub/