Intern, Translational Data Science; Project: Multiscale AI Tissue Representation and Biomarker Discovery

At Genmab, we are dedicated to building extra[not]ordinary® futures, together, by developing antibody products and groundbreaking, knock-your-socks-off KYSO antibody medicines® that change lives and the future of cancer treatment and serious diseases. We strive to create, champion and maintain a global workplace where individuals’ unique contributions are valued and drive innovative solutions to meet the needs of our patients, care partners, families and employees.

Our people are compassionate, candid, and purposeful, and our business is innovative and rooted in science. We believe that being proudly authentic and determined to be our best is essential to fulfilling our purpose. Yes, our work is incredibly serious and impactful, but we have big ambitions, bring a ton of care to pursuing them, and have a lot of fun while doing so.

Does this inspire you and feel like a fit? Then we would love to have you join us!

Are you excited about the intersection of AI and cancer immunology, looking to make a meaningful impact in the field of antibody therapeutics? Do you want to gain exposure to the strategic design of innovative antibody therapeutics? Keep reading!

We are looking for a highly motivated intern to join Genmab’s Translational Data Science team. You will work on two connected directions: multimodal models that integrate clinical imaging with molecular and genomic profiles, and an interpretable tissue-token codebook that translates morphology across radiology, whole-slide pathology, and molecular programs. The work will use public and commercial data to advance biomarker discovery and biologic hypothesis generation.

• Pillar 1 - Multimodal Biomarker Discovery:
  Design and validate models that integrate CT, MRI, or PET radiomics with molecular, genomic, and clinical data to identify meaningful, actionable biomarkers.
• Pillar 2 - Tissue Token Codebook Across Scales:
  Learn discrete tissue tokens in a shared semantic space spanning CT habitats, whole-slide imaging patterns, and molecular programs. Ground tokens in imaging, pathology, and omics evidence to create an interpretable biologic dictionary rather than a black-box embedding.
• Data Integration & Method Development:
  Use HEST-1k, TCGA, CPTAC-linked cohorts, radiogenomic imaging cohorts, and relevant internal resources. Explore CNNs, vision transformers, self-supervised learning, cross-attention, graph methods, and discrete representation learning.
• Feature Interpretation & Hypothesis Generation:
  Connect learned representations with genetic alterations, immune profiles, pathway activity, morphology, treatment response, and outcomes to support biomarker discovery and cross-study reuse.
• Model Validation, Generalization & Reproducibility:
  Conduct cross-validation, ablation studies, and external validation. Write clean, modular, well-documented code following reproducible research practices.
• Collaboration & Communication:
  Work closely with computational biologists, bioinformaticians, pathologists, imaging scientists, and clinicians. Present progress and findings in internal seminars and contribute to internal reports and potential publications.

• Currently enrolled in a Master’s degree program at a Dutch university in Computer Science, Artifical Intelligence or a related quantitative discipline.
• This internship must be a formal, mandatory component of your degree program required for graduation; we cannot consider candidates who have already graduated or who are seeking an extracurricular internship.
• Available for a minimum of 6 months.
• Proficiency in Python and deep learning frameworks such as PyTorch.
• Experience developing CNNs, transformers, self-supervised models, or multimodal architectures for medical imaging,…