AI-Driven Radiomics and Multimodal Biomarker Discovery Intern

Genmab • Plainsboro Township, NJ

49d • Hybrid

About The Position

As an AI-Driven Radiomics and Multimodal Biomarker Discovery Intern, you will join Genmab’s Translational Data Science team for a 10-week summer internship. You’ll work at the forefront of AI and computational oncology, developing next-generation deep learning frameworks that integrate radiomic features from clinical imaging with molecular and genomic profiles. Radiomics transforms medical imaging data (CT, MRI, PET) into high-dimensional quantitative descriptors capturing tumor morphology, texture, and spatial heterogeneity. Integrating these features with genomic and molecular datasets provides an unprecedented opportunity to identify multiscale biomarkers predictive of therapeutic outcomes. This internship will leverage both public datasets (e.g., TCGA-LUAD/LUSC, NLST) and commercial imaging and molecular resources, providing a rich landscape for developing generalizable multimodal models. You’ll apply CNNs, transformers, and self-supervised learning to bridge imaging and molecular data for biomarker discovery. This project aims to uncover novel, clinically actionable biomarkers to advance precision oncology. You’ll contribute to the design, implementation, and validation of multimodal AI architectures that generate new insights into tumor biology, treatment response, and patient stratification.

Requirements

• Currently pursuing a PhD or advanced Master’s degree in Computer Science, Data Science, Biomedical Engineering, Computational Biology, or a related quantitative discipline.
• Proficiency in Python and deep learning frameworks such as PyTorch.
• Demonstrated experience developing CNNs, transformers, or multimodal architectures for medical imaging, omics, or related AI applications.
• Experience with data management and distributed training frameworks (e.g., Weights & Biases).
• Familiarity with radiomics extraction libraries (e.g., PyRadiomics, MONAI).
• Knowledge of model interpretability tools (Grad-CAM, SHAP, feature attribution) and evaluation metrics for biomedical data.
• Analytical mindset, collaborative spirit, strong organizational skills, and proactive attitude.

Nice To Haves

• Strong understanding of or interest in cancer biology

Responsibilities

• Deep Learning Architecture Development: Design, train, and optimize advanced CNN-based and hybrid architectures (e.g., 3D CNNs, Vision Transformers, CNN -Transformer hybrids) to extract biologically meaningful radiomic representations.
• Multimodal Data Integration: Develop deep fusion models that combine imaging, molecular, and clinical data through cross-attention, late fusion, or graph-based techniques to enhance biomarker prediction and interpretability.
• Software Engineering & Reproducibility: Write clean, modular, and well-documented code following modern software engineering best practices.
• Feature Interpretation & Biomarker Discovery: Correlate learned features with genetic mutations (e.g., KRAS, EGFR, TP53), immune profiles, and clinical outcomes to identify interpretable and actionable biomarkers.
• Model Validation & Generalization: Conduct rigorous cross-validation, hyperparameter optimization, and external dataset validation to assess model robustness and reproducibility.
• Collaboration & Communication: Work closely with computational biologists, bioinformaticians, and clinicians. Present progress and findings in internal seminars and contribute to internal reports and potential publications.

What This Job Offers

• Career Level: Intern
• Number of Employees: 1,001-5,000 employees