Jupyter is one of the most influential tools in modern science, education, and data-driven work. Whether in classrooms, research labs, or industry, Jupyter notebooks have become a common way for people to think, compute, and work together.
In preparing for GW OSCON, our team had the opportunity to speak with Dr. Brian Granger, co-creator of Project Jupyter, longtime open-source leader, and current leader at Amazon Web Services, about how Jupyter came to be, why open source is so integral to scientific work, and what excites him most about the future of the field.
Why (and How) Jupyter Took Off
It’s easy to see that Jupyter has an extensive fan base. When asked what exactly made Juptyer so widely applicable across disciplines, Granger was quick to emphasize that its success didn’t come from one individual factor. Instead, Jupyter’s success emerged at the intersection of several tidal waves.
One reason was the rise of Python as a leading language for scientific computing, data science, machine learning, and AI. Another was the growth of open-source Python libraries like NumPy, Pandas, Matplotlib, and SymPy, which helped researchers solve real problems. Jupyter became valuable because it enabled all these tools to work together smoothly. As Granger put it plainly:
“Without all of these packages, Jupyter would not have been particularly useful.”
Granger also pointed out the importance of the open-source movement, the huge increase in available data, and fast progress in algorithms. Jupyter did not create these changes, but it was built to help people make the most of them.
Interactive Computing: a New Way of Thinking
Granger also highlighted two main design ideas that make Jupyter unique:
The first idea is interactive computing. Jupyter was not designed as a typical software engineering tool. Instead, it treats programming as a way to think, where people explore data, test ideas, and learn in real time. In this approach, code is part of the scientific and learning process, not just a tool.
“Jupyter is not for software engineering — it’s a tool for that tight loop of human thinking and looking at data, looking at a problem, and using programming in an interactive way.”
The second idea is what Granger calls ‘computational narrative.’ Jupyter notebooks combine live code, results, and explanatory text into a single document. That makes them not just tools for individual work, but powerful mediums for communication and collaboration. A notebook can show not only what someone did, but how and why they did it. Together, these ideas explain why Jupyter notebooks have become more than personal scratchpads. They’re artifacts that support teaching, teamwork, and shared discovery.
Reproducibility Without the Extra Burden
Reproducibility remains one of the most complicated challenges in science. While Granger was careful not to claim that Jupyter “solves” it, one could argue that tools like Jupyter notebooks make reproducibility significantly easier. Full reproducibility depends on many factors: software environments, dependencies, data access, and long-term maintenance.
“If reproducibility is an add-on, is something that you do later… it’s just never gonna get done,” he said.
Granger said that Jupyter helps make reproducibility a natural part of the process, not just an afterthought. When researchers use notebooks, they often create work that is easier to share and more transparent than older methods. Reproducibility becomes a regular part of their work, not just an extra step at the end.
“Jupyter gives people a tool that they can just work in, that’s pleasant to just work in,” Granger explained. “At the end of that, they look down and realized, oh, I’ve got a set of notebooks that I can share with people.”
The Many Hats of Open Source
Granger’s career has spanned academia, open-source leadership, and industry—a path that may seem daunting to students considering their own futures. But in his view, open source is precisely what makes these transitions possible.
Open source drives innovation in universities, tech companies, and cloud platforms like AWS. Many important advances in AI and data science, including Jupyter, PyTorch, and LangChain, come from open collaboration.
For students, Granger emphasized that open-source contributions are an extraordinary form of professional training. While small contributions matter, working on complex, long-term problems in open source offers invaluable experience in collaboration, problem-solving, and leadership.
Sustainability, AI, and What Comes Next
Looking to the future, Granger and many others are especially concerned about keeping open source sustainable. Many popular projects serve millions but depend on only a few maintainers, which can be discouraging and hard to keep up with. Open source relies on community support, and once a project is abandoned, it is difficult to revive it.
Granger is cautiously optimistic about AI-assisted software development. If used wisely, AI tools could help small teams maintain and even improve large, long-running projects by rewriting old code, making it easier to manage, and opening up new possibilities.
With AI-assisted development, “we’re seeing ways that… make it plausible for a small number of people working on and maintaining a large-scale open source project to get a lot more done.”
For Granger, using AI even in his personal life isn't about simply working faster. It’s to do better, more impactful work and to rethink old assumptions about how software is built and sustained.
We at the OSPO are excited to welcome Brian Granger to GW OSCON, where he will share more about these topics in his keynote this March. His experience in technology, community, and long-term planning shows why open source is so important in research, education, and more. Whether you are a student starting out, a researcher focused on reproducibility, or a contributor looking to the future, Granger’s work reminds us that tools are important, but communities built with them are even more so.