Modern computational methods are increasingly becoming an essential tool throughout physics. However, their practical use within Physics is often built upon combined knowledge of computational methods and physics that are taught separately. We present a course that provides realistic, contemporary examples of how computational methods apply to physics research, and deliver this content via interactive Jupyter notebooks. This course, titled "Computational Data Science in Physics," was delivered in several different modalities from 2021 to 2023, ranging from online modules on the MITx Online platform (using Open edX), to a full semester, graduate-level course at MIT. For the online modules, we developed interactive problem graders for coding problems and organized content to promote active learning (e.g., lecture videos intermixed with exercises). Each module culminated in a Final Project, where students applied what they had practiced in previous lessons towards a recent (Nobel prize winning) data set (e.g., LIGO and LHC data). Importantly, we ensured that notebooks were accessible to learners in several formats in order to broaden the modes with which learners could engage with the content.
