Machine Learning Foundations and Biomedical Discovery
Our overarching goal is to lay the foundations of AI for science, biology, and medicine, creating systems that deepen scientific understanding, accelerate therapeutic discovery, and increasingly learn from data, knowledge, and experimentation.
We develop foundational advances in artificial intelligence and machine learning, with emphasis on models and systems that are grounded in knowledge. Our work includes pre-trained, self-supervised, multimodal, and agentic AI models trained that can reason and empower scientific discovery across molecular, cellular, tissue, and patient scales.
Open AI Scientists Initiative | Join Us
AI scientists are autonomous AI systems that reason, hypothesize, and experiment alongside human researchers. Through our Open AI Scientists Initiative, we develop and study AI scientists that integrate research and clinical data with scientific tools to generate new hypotheses, prioritize experiments, and produce scientific insight. We evaluate these systems through discovery loops that connect AI reasoning with experiments in biological and clinical labs.
AI for Science | Scientific Discovery and Therapeutic Science
Science is entering a new era in which AI can do more than analyze data: it can help generate hypotheses, connect disparate evidence, guide experiments, and accelerate discovery. Biological systems are deeply interconnected across scales, from genes and proteins to cells, tissues, and whole organisms. Our research develops AI methods that learn across these levels of organization to uncover biological principles, model disease mechanisms, and advance the design of therapeutics and other interventions.
AI for Medicine | Individualized Diagnosis and Treatment
Human health can now be measured with unprecedented precision through genomic data, cellular atlases, molecular profiles, clinical records, imaging, and treatment histories. The central challenge is to integrate and reason over these heterogeneous data to support diagnosis, prognosis, and treatment selection. Our research builds AI systems that combine patient data with biomedical knowledge to enable more precise, evidence-based, and individualized medicine, with the goal of matching each patient to the right treatment and improving consistency between clinical outcomes and biological insight.