On the final morning of the National Workshop on Data Science Education, BIDS Executive Director, Kirstie Whitaker, presented a panel that exemplified what makes UC Berkeley’s data science community a driving force behind the future of open, collaborative infrastructure. Each speaker, deeply involved in projects like Jupyter, GeoJupyter, Jupyter Book, and Scientific Python, shared insights into how their work is shaping a broader ecosystem grounded in accessibility, reproducibility, and community.
Fernando Pérez, BIDS Faculty Director and co-founder of Project Jupyter, kicked off the session by emphasizing the foundational role of community:
Jupyter is more than just a notebook or a tool. Instead, it’s a community of people working to make science more efficient, better connected, and innovative. — Fernando Pérez
This message highlighted the values that drive technical innovation at BIDS. Pérez demonstrated a prototype of a chatbot interface layered over Jupyter Notebooks, showing how AI could lower the barrier to entry for interacting with code. Rather than replacing human insight, the demo illustrated how conversational tools can support exploration, helping newcomers and experienced users engage intuitively with data.
Following Pérez, DSE Senior Program Director Ciera Martinez introduced GeoJupyter, a project that aims to transform how people interact with geospatial data. With a focus on equity and social impact, Martinez highlighted how tools that display maps can shape more than just workflows.
Who controls the map controls the narrative. It’s not just a technical issue. It’s a social one. Geospatial data are critical tools people rely on every day to make decisions about their communities, livelihoods, and futures. And yet, access, control, and interpretation of this data remain deeply uneven. Our goal is to build a collaborative ecosystem that supports users across the entire spectrum, from those just beginning their data journey to those pushing the boundaries of multidimensional analysis. — Ciera Martinez
Photo: Ciera Martinez, introducing the GeoJupyter project at the National Workshop on Data Science Education.
GeoJupyter’s browser-based interface enables a wide range of users to engage directly with map-based data, facilitating more informed decisions about land use, policy, or resource management. Panelist Chris Holdgraf then shared updates on Jupyter Book, a tool that helps authors weave together code, narrative, and output into rich, reproducible, multi-document websites. Built on MyST Markdown, Jupyter Book treats content as structured data, making it easier to preview, cross-link, and reuse materials. The platform’s modular design allows teams to generate content once and publish it across multiple formats without duplication. More than just a publishing tool, Jupyter Book supports collaborative and interactive learning environments. Driven by an international community of contributors, Jupyter Book helps close the gap between reading and writing computational narratives.
Sean Morris, representing his colleague Balaji Alwar (UC Berkeley DataHub/CDSS), also highlighted accessibility tools, particularly the JupyterLab A11y Checker, an extension that helps Jupyter Notebook authors identify and resolve issues like low color contrast or missing alt text. Now deployed on UC Berkeley’s DataHub, the checker runs directly in the browser and can even suggest or implement fixes in real time. Designed with community feedback, the A11y Checker is helping bring Jupyter notebooks up to AA accessibility standards, ensuring that the Jupyter ecosystem continues to grow its inclusivity as it evolves.
Closing the panel was astronomer and Scientific Python contributor Brigitta Sipőcz. Drawing from her work across Jupyter, Jupyter Book, and the broader open science community, Sipőcz underscored the importance of transparency, testing, and user contribution:
Executable, tested tutorials should be treated with the same care as the software itself. By applying continuous integration and open communication practices to scientific documentation, we make research more reproducible, dependable, and inclusive. When users feel empowered to report issues and contribute, they become collaborators, not just consumers. — Brigitta Sipőcz
Sipőcz’s remarks highlighted the often invisible labor behind reliable documentation and emphasized how even small contributions strengthen the open network for everyone.
From foundational infrastructure to user-first design and inclusive documentation practices, the Friday morning panel offered a powerful snapshot of what’s next for open science. Our colleagues aren’t just building tools; they’re building systems of trust, care, and collaboration.
Photo: (from left) Eric Van Dusen, Fernando Pérez, Ciera Martinez, Sean Morris, Brigitta Sipőcz, Chris Holdgraf (back), Kirstie Whitaker
For more details on the initiatives discussed, explore the speakers’ presentation slides below:
- Jupyter Book 2.0 and the MyST Engine (Chris Holdgraf)
- User facing tutorials as code reproducible and reliable tutorials with CI/CD (Brigitta Sipőcz)
- GeoJupyter (Ciera Martinez)
- JupyterLab-a11y-checker Intro & Demo (Sean Morris, representing his colleague Balaji Alwar)
We invite you to explore, contribute, and help shape the future of open science.