TUM Klinikum Rechts der Isar
Technische Universität München
Wissenschaft und Forschung
Postdoc/PhD in Computer Science – Multimodal Models
Veröffentlicht: 27.05.2025

Full- or Part-Time | Limited Contract (24 Months) | Starting June 1, 2025

 

Are you passionate about building state-of-the-art AI that positively impacts cancer diagnostics and treatment? We are seeking a highly motivated Computer Scientist to paritcipate in the development of a novel Multimodal Foundation Model for Cancer Imaging and Biomarker Discovery. This is a unique opportunity to work with large-scale, diverse datasets and contribute to a foundational AI technology with high translational potential in oncology.

 

Your tasks:

  • You will play an integral role designing and implementing a cutting-edge foundation model that integrates high-dimensional medical imaging data (CT, MRI) with unstructured radiological text reports.
  • You will develop and implement robust pipelines for the curation, integration, and preprocessing of heterogeneous data from diverse sources, including clinical PACS systems and large-scale research cohorts (e.g., NAKO, UK Biobank, TCIA). This includes tasks like image stitching, harmonization, and de-identification. All data sources are already secured, so development starts at day zero. 
  • Implement and refine innovative learning strategies, including contrastive learning (anatomical, multi-view, pathological contrasts) and weakly supervised approaches, to train a model for robust feature extraction and generalizability. You'll explore novel cross-attention mechanisms for effective image-text fusion
  • Your work will enable the extraction of generalizable, image-based biomarkers from the learned representations, aiming to improve clinical decision-making in oncology.
  • Develop and optimize data loading and training infrastructure (PyTorch, MONAI) for efficient handling of large 3D medical datasets and high-performance GPU clusters, including modality-specific data augmentation and adaptive sampling algorithms.
  • Work within a dynamic, interdisciplinary team of clinicians and AI researchers, and disseminate your findings through high-impact publications and conference presentations.

 

Your profile:

  • An excellent Master's degree (for PhD position) or PhD (for Postdoc position) in Computer Science, Artificial Intelligence, Data Science, or a related field.
  • Proven, advanced programming skills, particularly in Python, and experience with deep learning frameworks (PyTorch highly preferred).
  • Demonstrable experience in developing and training deep learning models, ideally with a focus on medical image analysis (e.g. with MONAI). 
  • A passion for tackling complex scientific challenges, a creative mindset for developing novel solutions, and a meticulous approach to research and validation.
  • You are passionate about open science and open source. 

 

What we offer:

  • The opportunity to work on cutting-edge research projects with high societal impact.
  • Access to unique, large-scale medical datasets and high-performance computing infrastructure (including NVIDIA A100/H100 GPUs).
  • An interdisciplinary and international working environment with close collaboration between AI experts and leading clinicians.
  • Funding for conference travel and publications.
  • EGYM well pass, corporate benefits and discounts (e.g. Käfer), cafeteria, sports and cultural offers
  • Free use of the library through a branch of the Munich City Library located in the building 
  • Working in the heart of Munich at Max-Weber-Platz with very good accessibility by public transport such as the subway, S-Bahn or tram
  • Company pension plan through the Federal and State Pension Institute (VBL)

 

We look forward to your application!

 

Contact: PD Dr. Lisa Adams | 089 / 4140 –1084 | Institut für diagnostische und interventionelle Radiologie

 

Please submit your complete application documents by e-mail including:

  • reference number 25_05_041.

     

Institut für diagnostische und interventionelle Radiologie

 

TUM Klinikum
Rechts der Isar
Ismaninger Straße 22
81675 München

E-Mail: keno.bressem@tum.de

 

 

If the candidates’ suitability for the position in question is equal, severely disabled applicants shall be given preference. Interview-related costs can, unfortunately, not be reimbursed.