Postdoctoral Fellow, Scientific Machine Learning for Oncology Digital Twins

A healthier future. It’s what drives us to innovate. To continuously advance science and ensure everyone has access to the healthcare they need today and for generations to come. Creating a world where we all have more time with the people we love. That’s what makes us Genentech.

Genentech’s Clinical Pharmacology Modeling & Simulation group seeks a postdoctoral fellow (or artificial intelligence (AI) residency) to advance scientific machine learning for clinical oncology. The project builds a hybrid framework that couples an existing quantitative systems pharmacology (QSP) model of cancer immunotherapy with graph neural networks trained on spatial single-cell tumor microenvironment (TME) data from non-small cell lung cancer (NSCLC). Using high-dimensional datasets, you will learn bi-directional links between local TME interactions and systemic immunity, create digital twins to test treatment hypotheses, and inform patient-selection strategies. You will integrate spatial transcriptomics with clinico-genomic/response data, design training over baseline, on-treatment and post-treatment time points, and model treatment effects with both mechanistic and data-driven components. You will work with Clinical Pharmacology, Research Oncology, Biomarker Development and AI/ML peers, publish work not tied to a specific program, and help disseminate findings internally and externally. This is a high-impact role with strong mentorship and opportunities for first-author publications. The appointment is from 2 to 4 years.

The Opportunity:

In addition to developing a hybrid mechanistic–ML framework for the comprehensive assessment of immunotherapy and combination strategies, the fellow will compile, harmonize, and quality-check all required data (e.g., spatial/single-cell, genomics, clinical outcomes); set up and maintain the end-to-end computational stack (data processing, model training/inference, simulation, visualization) using reproducible practices (Git version control, automated tests/CI); work closely with modeling and AI/ML colleagues while collaborating across Oncology, Biomarker, and Clinical teams; and deepen expertise in tumor immunology and clinical aspects of anti-cancer treatments as needed. 

Other responsibilities include:

• Present results in accessible terms, and actionable recommendations, at cross-functional teams, department meetings, and review committees.

• Prepare posters and talks for internal seminars and external conferences.

• Write manuscripts (lead and co-author) summarizing methods, results, and insights.

• Adapt and thrive in a collaborative, interactive, and team-oriented environment.

Who You Are:

• Ph.D. in physics, applied mathematics, statistics, computer or computational sciences, or in an engineering field such as biomedical informatics, or a related discipline.

• Required experience in developing deep-learning systems; graph neural networks & representation learning (VAEs/transformers) preferred.

• Experience in statistical modeling & ML for high-dimensional data (training, evaluation, uncertainty) is required.

• A publication record of substantial/influential work is expected.

Preferred Qualifications:

• Experience with ODE-based systems modeling (especially of biological systems), such as Quantitative Systems Pharmacology (QSP), would be highly preferable. Knowledge of neural-ODE or Universal Differential Equations is a plus

• Proficiency in Python or Julia (Python preferred) & modern ML tools (PyTorch, JAX or TensorFlow); hands-on knowledge of Matlab is a plus; modern software practices (Git version control, automated tests/CI).

• Experience with single-cell and/or spatial transcriptomics & integration with clinical/response data, or strong motivation to learn.

• Familiarity with causal inference or generative modeling for virtual populations/digital twins advantageous.

• Strong communication and interpersonal skills, enthusiasm to contribute to a multidisciplinary environment, and a drive to solve challenging scientific problems.

• We are looking for creative, resourceful, and intellectually curious individuals who