Overview In this role you advance longitudinal multimodal foundation models for healthcare, integrating imaging, EHR, genomics, and other data to model disease progression and treatment response. You join an applied research team that bridges publication-quality work with practical software and open-source contributions, collaborating with healthcare organizations and industry partners to translate research into real-world AI solutions. You will shape foundational architectures, training strategies, and benchmarks that propel healthcare AI forward, while contributing across multidisciplinary teams. This position offers meaningful impact through scalable AI models that support precision medicine and患者
Compensation / Benefits- equity and benefits package
- base salary plus potential for equity
- competitive compensation
- opportunities for career growth
Responsibilities- Research longitudinal multimodal foundation models across heterogeneous healthcare data
- Develop novel model architectures and training strategies for disease progression and temporal reasoning
- Build large-scale datasets, benchmarks, and open-source foundation models
- Collaborate with Healthcare AI, BioNeMo, and NVIDIA teams to integrate imaging and molecular data
- Partner with healthcare institutions and industry partners to translate research into software and workflows
- Publish research, contribute to open-source software, and help define future directions of healthcare foundation models
Key requirements- PhD in a quantitative field or equivalent experience
- 8+ years of industry experience in medical AI research
- Experience building multimodal foundation models with longitudinal modeling or disease progression focus
- Proficiency in PyTorch and large-scale model training
- Strong software engineering and experimental skills for scalable, reproducible pipelines
- Excellent communication and cross-disciplinary collaboration abilities
- cross-functional collaboration
- communication
- autonomy
- PyTorch
- large-scale foundation model training
- distributed GPU training
