Research Scientist, Medical World Models

What you'll do

• Design and train longitudinal forecasting models and multimodal health-state representations.
• Own model training and evaluation code, and deliver documented, reproducible pretrained models to product teams.
• Validate models responsibly, communicate limitations, collaborate with clinical and engineering teams, and publish research.

What they ask for

• PhD in machine learning, computer science, biomedical engineering, or a related field, or MS/BS with 5+ years building and evaluating ML models on real data.
• Depth in at least one area: temporal/longitudinal modelling, medical or multimodal foundation-model pretraining, or multimodal fusion of imaging with tabular, EHR, or biomarker data.
• Strong Python and PyTorch; experience training at scale and writing code others build on.
• Rigorous evaluation skills, plus a publication record at top ML or medical-imaging venues or equivalent production research output.

Nice to have

• Experience with world models, latent dynamics, model-based RL, or counterfactual/causal inference on observational health data.
• Experience with longitudinal cohorts, EHR/lab time series, clinical foundation models, cloud ML infrastructure, or regulated healthcare environments.

Tools

Python, PyTorch, GPU, AWS, Databricks, LLMs

AI focus

Develop multimodal medical foundation models that represent current health, forecast longitudinal health changes, and provide pretrained encoders and forecasters for clinical applications.