Fully Funded PhD Position in Differential Privacy for Medical Deep Learning at RWTH Aachen University, Germany

Applications are invited for a fully funded PhD position in Differential Privacy for Medical Deep Learning at the Reliable Medical AI group, part of the Lab for AI in Medicine at RWTH Aachen University and University Hospital RWTH Aachen, Germany. Funded by the German Research Foundation (DFG), this position offers doctoral candidates an exceptional environment with cutting-edge computing infrastructure to conduct foundational research in trustworthy AI.

Overview

Designation

PhD Student / Doctoral Researcher (m/f/d) – Full-time

Research Area

The research is centered on Differential Privacy (DP) for Medical Deep Learning. The project focuses on medical foundation models, specifically examining the theoretical and empirical mechanisms by which differentially private stochastic gradient descent (DP-SGD) affects model utility across diverse patient subgroups, with the ultimate goal of developing novel DP finetuning algorithms that preserve high diagnostic accuracy for underrepresented groups.

Location

Aachen, North Rhine-Westphalia, Germany (Reliable Medical AI group, Lab for AI in Medicine, RWTH Aachen University).

Scholarship & Project Description

Medical foundation models are typically pretrained on large image collections and finetuned on hospital-specific patient data. While differentially private (DP) finetuning guarantees individual privacy protection, it often compromises diagnostic performance unevenly across different patient cohorts. This research aims to understand these mechanisms and create equitable, high-performing DP finetuning methodologies.

Key aspects of the doctoral project include:

• Analyzing how gradient clipping, noise addition, and subsampling interact with pretrained representations using theoretical and empirical approaches.
• Developing reliability-aware objectives and controlled adaptation strategies to enhance privacy-preserving medical models.
• Publishing findings in premier peer-reviewed venues including NeurIPS, ICLR, ICML, CVPR, Nature Portfolio, and MICCAI.
• Access to state-of-the-art computational infrastructure featuring a GPU cluster equipped with 30 NVIDIA H100 and 16 RTX 6000 Blackwell GPUs, along with curated, large-scale medical imaging datasets.

Eligibility / Qualifications

Applicants must meet the following criteria:

• A Master’s degree in Computer Science, Electrical Engineering, or a closely related discipline.
• A rigorous theoretical and technical foundation in machine learning and deep learning.
• Proficiency in Python and deep learning frameworks (e.g., PyTorch).
• Strong written and verbal English communication skills.

Preferred Qualifications:

• Prior research publications in high-impact machine learning or computer vision conferences/journals.
• Prior theoretical or practical background in differential privacy.
• Experience utilizing agentic AI tools (e.g., Claude Code) is a welcome plus.

How to Apply

Interested candidates should submit their complete application package by email directly to Dr.-Ing. Dr. rer. medic. Soroosh Tayebi Arasteh at stayebiarast@ukaachen.de.

Please compile the following documents into a single PDF file:

1. Curriculum Vitae (maximum 3 pages)
2. Cover Letter explaining your motivation and research alignment (maximum 1 page)
3. Master’s degree transcript and academic records
4. Three (3) letters of recommendation

Last Date for Apply

Applications are reviewed on a rolling basis until the position is filled. Early submission is strongly encouraged.

Apply Link

For more details or to submit your application, consult the official page or send your materials directly via email:

Visit the Official Job Announcement Page