Scaling Instructional Workflows in Data Science Education using JupyterHub and Otter-Grader

Volume 16, Issue 2 (November 2025), pp. 2–4

Download PDF

@article{jocse-16-2-1,
  author={Sai Annapragada},
  title={Scaling Instructional Workflows in Data Science Education using JupyterHub and Otter-Grader},
  journal={The Journal of Computational Science Education},
  year=2025,
  month=nov,
  volume=16,
  issue=2,
  pages={2--4},
  doi={https://doi.org/10.22369/issn.2153-4136/16/2/1}
}

At the University of California, Merced (UCM), an instructional workflow was adopted to support the teaching of data science at scale. This workflow integrated JupyterHub, Otter-Grader, and GitHub to facilitate browser-based notebook execution, simplify assignment distribution, and automate grading. Initially built around a shared-folder model—--where instructors placed course materials in a shared---readwrite directory that automatically appeared as a read-only shared directory for all students—--the system transitioned to a GitHub-based setup using nbgitpuller. This shift allowed instructors to distribute assignments and course materials through direct links, removing the need for students to navigate the shared folder manually. By doing this, the need for admin privileges was removed, reducing the risk of accidental deletion of course content from the shared read-write folder. This paper presents our instructional strategy, key challenges addressed, implementation experience, and insights for educational institutions seeking to adopt similar models.