He teaches in the Biomedical Informatics (BMI) graduate degree program, holding the positions of Professor of Medicine, Associate Dean of Research, and Chief Data Scientist at Stanford Health Care.

[22][23] His team released the foundation models that are trained using de-identified, longitudinal electronic health records along with a benchmarking dataset with manually verified labels for open comparison of technical advances.

[37] A large portion of his research focused on validating the core informatics methods to enable such use of aggregate patient data, leading to the Green Button initiative.

[38][39] He later operated a bedside consultation service in the USA that provides a clinician with an on-demand summary of similar patients in terms of the treatment choices made and observed outcomes.

[47] His work demonstrated that using simple term recognition methods results in little or no impact on accuracy for multiple clinical tasks with the advantage of scaling to very large datasets.