On October 16, a range of professionals, students, and open source advocates gathered for a panel discussion led by Kirstie Whitaker, Berkeley Institute for Data Science (BIDS) Executive Director, at the AI Futures Lab. She discussed the new membership plan with 2i2c (International Interactive Computing Collaboration), walking through their origins and upcoming plans with Chris Holdgraf, 2i2c Executive Director, and Yuvaraj (Yuvi), 2i2c Tech Lead and Co-founder. Their discussion focused on a shared mission: building sustainable, inclusive, and community-driven infrastructure for open research.
Together at this Open Source Program Office (OSPO) event, the speakers explored how open source tools can bridge disciplinary divides, empower educators and researchers, and foster the ecosystems that keep scientific collaboration thriving.
2i2c’s goal has been to connect two often separate worlds, open source developers and academic researchers, striving to create efficient collaboration. The organization’s model reinvests community-driven computational infrastructure back into open source projects, creating what Chris described as a “virtuous cycle” where improvements to one project are reinvested to benefit the wider open source community.
While 2i2c’s mission centers on collective progress, its roots lie in the individual paths of those who built it. Both Chris and Yuvi reflected on how their early experiences in research and education shaped their commitment to open source work.
Growing up in southern India, Yuvi shared how his early experiences with a cultural memory rooted in colonial history taught people around him to be wary of things labeled as “free,” and led to a strong open source movement in India. After engaging in projects with Gnome and Wikimedia, Yuvi spent a short time in the startup world before committing to nonprofits and open communities. Chris took a different path, which began during his neuroscience PhD program at UC Berkeley, where he and Kirstie met. He joined the early wave of researchers drawn to Python as an alternative to MATLAB, and realized that he cared more about the research ecosystem than any single domain. This led him to support infrastructure that empowers researchers broadly, a principle that would later inspire 2i2c’s founding.
Photo: (from left) Kirstie Whitaker, Yuvaraj (Yuvi), Chris Holdgraf talking about their personal open source journeys
While open source work often measures visible output, the speakers emphasized the importance of sustaining the broader ecosystem as well. Chris noted how open source projects function as independent communities, not just company projects, and described the challenge of supporting these projects while also managing the priorities and responsibilities of their own organization.
At 2i2c, this crystallized into the idea of balancing direct and foundational contributions. Chris explained that while direct contributions are work specific to the immediate needs of their partner communities, foundational contributions strengthen the health of the projects themselves, through maintenance, governance, and documentation. Since its founding in 2020, they’ve explored ways for their team to contribute upstream to the broader projects they depend on. By documenting these potential approaches, they hope to provide transparency and guidance to other organizations who can learn and adopt similar practices. Yuvi expanded on this, noting that even open, collaborative spaces need intentional structure to remain inclusive. He drew from feminist frameworks, noting that the tyranny of structurelessness can reproduce existing hierarchies. To address this, 2i2c is experimenting with practical approaches: retaining newer contributors through code review, distributing release responsibility, and shifting mentorship to sponsorship. While many participants are technically skilled, the key lies in empowering them with the right support and providing space to contribute meaningfully.
After working with different community partnerships, 2i2c is formalizing its model through a membership program with BIDS as its first premier member. The goal is to turn the hodgepodge of past collaborations into a more standardized framework – a sustainable, repeatable model – rather than one-off’s. As a result, 2i2c has more intentional engagement with its members which allows them to define what support looks like across disciplines and to drive investment back into those spaces.
Kirstie emphasized how this aligned with earlier points about structure and support: “By maintainers, for maintainers.” She highlighted Carl Boettiger, who developed code for running LLMs on national research platforms. Traditionally, sharing this work would require him to take on a new passion project to support users. 2i2c’s approach ensures he isn’t burdened alone. Contributions are shared upstream, work and credit are distributed appropriately, and funding flows to the right organizations and people, building a scientific open source ecosystem that is both sustainable and inclusive. Yuvi elaborated on this idea, noting that co-creation is essential: “Build and they come is a lie, for the most part. You need to build together with the people who will actually use it so it meets their needs.” By connecting people with different expertise and creating collaborative spaces, 2i2c ensures that improvements and resources benefit the entire community.
Looking forward, Chris described 2i2c’s ambition to continue growing a network of community hubs, through their managed JupyterHub service, that share roughly 80% of the same building blocks. While workflows vary, the shared tools allow for deployments and improvements to be pushed across the entire network while leaving the remaining 20% for local customization. The network also aims to facilitate collaboration by making the needs of different groups transparent. Multiple groups often face similar hurdles, and this network would allow them to pool and coordinate resources toward shared goals. This method ensures that improvement benefits multiple groups while maintaining sustainable and inclusive practices, circling back to how the membership program creates upstream contributions to support the broader ecosystem.
The conversation closed with thoughtful questions from the audience, ranging from the specifics of their code review practices to how the organization supports community colleges and other institutions. Afterward, attendees lingered to chat, exchange ideas, and reflect on the possibilities that open, shared infrastructure can bring to research. Min Ragan-Kelley, BIDS Senior Open Infrastructure Architect, whose work is directly impacted by the new membership, noted: “BIDS joining 2i2c helps us deliver positive impact across scientific and educational communities by contributing to shared open infrastructure, such as JupyterHub and mybinder.org.”
Photos: Guests gather after the fireside chat for more conversation and refreshments.
This event marked the start of a new partnership between BIDS and 2i2c, reinforcing a shared commitment to building sustainable, equitable, and community-driven research infrastructure. The collaboration reflects both organizations’ belief that open science doesn’t just happen in code, but through the relationships, conversations, and collective effort that sustain it.
For more on this partnership, read Berkeley Institute for Data Science partners with 2i2c on open source infrastructure, an article from the UC Berkeley College of Computing, Data Science, and Society (CDSS).
To hear more about open source initiatives at BIDS, join the BIDS mailing lists!