Sustaining OpenTelemetry: What a 10-week contributor cohort actually looks like

A follow-up to our earlier post: “Sustaining OpenTelemetry: Moving from Dependency Management to Stewardship” *** In April 2026, the Cloud Native Computing Foundation (CNCF), the OpenTelemetry (OTel) project, and Bloomberg’s Open Source Program Office came together to run a 10-week mentorship cohort together. We called it a “structured contributor pipeline.” We weren’t sure exactly what we’d learn. Now that the cohort has wrapped, we have results – and they’re worth sharing. What actually happened 48 Bloomberg engineers participated in the initiative. Most had never contributed to open source before. Some came in with deep OTel experience. Here’s what the cohort contributed to the project over the course of 10 weeks: - 118 pull requests (PRs) were submitted to the OpenTelemetry project. - 70 were merged – accepted into the global codebase by the worldwide maintainer community. - Contributions spanned 11 repositories: the Collector, Python SDK, Go SDK, Erlang SDK, Rust SDK, opentelemetry.io, the OTel Demo, instrumentation libraries, the Ecosystem Explorer, and more. - 842 volunteer hours were logged by participants During this timeframe, the OpenTelemetry project also reached its own milestone: it achieved fully graduated status within the CNCF, placing it alongside Kubernetes as one of the most trusted open source projects in the world. What the contributions actually looked like The work done was consequential. The Bloomberg engineers spent time identifying and working on crucial project maintenance – the kind that matters. Florian Bourgey of Bloomberg’s Quant Research team in the firm’s Office of the CTO systematically worked through a renaming campaign in the OTel Demo repository, submitting 23 pull requests – all of which were merged – that standardized telemetry attribute naming across the entire project (see issue #3267). By the end of the 10-week initiative, he had become the #2 most prolific contributor to the OTel Demo. This is exactly the kind of high-leverage, high-toil work for which open source project maintainers rarely have bandwidth. Engineers Thomas Baldwin, Mikiyas Bokan, and Larry Zebaze of Bloomberg’s Public Cloud Observability engineering team brought a real problem from their work into the open source community: the OTel Collector’s static secret storage required full restarts when credentials rotated. Seeing an opportunity to solve this challenge and contribute the solution upstream, they filed a community issue and built a multi-cloud credential rotation extension for the Collector — covering AWS, GCP, and Azure. Their work is still in review by the project maintainers, but will outlast the cohort in a meaningful way that benefits the greater OTel user base. Others contributed bug fixes to Python instrumentation, documentation improvements across opentelemetry.io, reference applications in JavaScript and Python (FastAPI), C++ code health fixes, Go SDK improvements, and tooling that will help future contributors work across multi-repo Python environments. Not all of their contributions were code. Several participants spent their volunteer hours reading spec documents, triaging issues, reviewing PRs they didn’t author, and asking questions that led other contributors to the right answers. That work matters too. What the mentors observed The OpenTelemetry community brought extraordinary generosity to this cohort – seven maintainers from around the globe who volunteered their own time week after week to answer questions, review code, and unblock stuck contributors. Each week brought a different deep dive to help educate the volunteers about the OTel project: semantic conventions and Weaver, the OTel Demo, GenAI observability, the Collector architecture, the path from contributor to maintainer, and more. Juraci Paixão Kröhling, a member of the OpenTelemetry Governance Committee, reflected on the experience at the final session: “This is one of the most successful mentorship programs I have seen at CNCF.” Liudmila Molkova of Grafana Labs, who led our GenAI observability session, offered advice that resonated across the cohort: “Find something applicable to your actual work that also helps your main job – otherwise it’s not sustainable.” She also reframed how contributors should think about their impact: “Reviews are more important than contributions. If you consistently provide comments, I’d be happy to promote you to Approver.” Kemal Akkoyun, who led a deep dive on contributing to the OTel Collector, gave contributors a practical framework for getting work landed: “Signal intent before you write code – open an issue, describe the problem clearly, get alignment before spending time on a PR.” And for those just getting started: “Start small. A focused, well-scoped PR with a clear problem description moves faster than a large one. Small PRs give you a ‘dopamine hit’ and help you understand the process.” Martin Thwaites, who joined for the semantic conventions session, reminded participants that contribution takes many forms: “Share your experience, not just opinions – practical feedback is the most valuable contribution to these standards. Lurking at meetings counts as contributing.” Pierre Tessier tracked every cohort PR closely and celebrated individual contributor journeys – including noting that Florian Bourgey had gone from his first PR to the second-largest human contributor to the OTel Demo in just a few weeks. Severin Neumann closed the series with a message to participants and mentors alike: “Close what can be closed, merge what can be merged – and for everything else, keep the conversation going.” That observation carries weight. CNCF runs programs across hundreds of projects. What made this one work, we believe, is its structure: not a hackathon, not a grant, not a “contribute to our project” call-to-action – but a weekly cadence, a real onboarding path, and experienced guides who stayed engaged for the full 10 weeks. The internal team that made it work Behind the cohort’s weekly rhythm was a small organizing team within Bloomberg who designed the structure, tracked every participant, and kept the momentum going week after week. Engineer Devpriya Dave served as team lead and mentor – facilitating breakout rooms, coordinating with the maintainers, and making her own upstream contributions to opentelemetry.io along the way. Yoga Ramalingam and Stephen Black of the company’s Telemetry Infrastructure engineering team brought deep telemetry infrastructure expertise to every session, helping participants navigate development environments and connecting cohort learnings back to Bloomberg’s own use of OpenTelemetry. Alyssa Wright designed the series model, ran participant operations, and managed the partnership with the CNCF and the OTel maintainer community. Together, they created the conditions for 48 engineers to succeed — and for seven external maintainers to feel supported enough to take their time to engage deeply with the cohort week in and week out. Our sustainability model for open source simply doesn’t work without this layer. External mentors provide the project expertise. Internal shepherds provide the operational infrastructure, the participant care, and the bridge between a company’s engineering culture and the open source community it depends on. At Bloomberg, that infrastructure also includes a close partnership between the Open Source Program Office and our Corporate Philanthropy team, building on existing systems for volunteering and charitable giving, to make open source contributions easier to support, track, and recognize. That partnership is what makes the Bloomberg Sustaining Open Source Series more than a contributor cohort. It connects the practical work of upstream contribution to a broader model for open source stewardship: foundation partnerships, maintainer mentorship, employee volunteering, and philanthropy programs that recognize open source contribution as meaningful service. Bloomberg is an open source first company – contributing upstream isn’t a stretch, it’s how we build. Our approach simply gives more Bloomberg engineers a supported path to do it. What we learned A few things stood out that the partnership didn’t fully anticipate: The bottleneck is issue discovery, not motivation. Participants consistently told us they wanted to contribute but didn’t know where to start. The most effective thing we did was maintain a curated list of right-sized, available issues – and have the mentors help participants scope and claim them in real time. Maintenance work is genuinely satisfying. The contributors who logged the most hours and submitted the most PRs were often working on unglamorous, but meaningful, tasks: renaming attributes, migrating feature gates, fixing CI workflows, improving documentation. When contributors understand why that work matters, they engage with it enthusiastically. Cross-company collaboration compounds. The cohort included engineers from different Bloomberg teams who had never worked together. Several formed working groups that extended beyond this cohort. One team’s breakout room conversation turned into a multi-month engineering project. 94% of respondents to our post-cohort survey said they feel more confident contributing to open source now than before they started. 88% said they plan to keep contributing to open source. Clearly, the pipeline doesn’t stop when the cohort does. What this model suggests for others This model works because it’s built on partnership: internal shepherds who invest in their people, and open source project maintainers who open the door for them to contribute. The trust between those two sides is what makes the structure hold. For companies, that matters. Open source sustainability requires more than dependency management, sponsorship, or a call for employees to “go contribute.” It requires a practical operating model. Organizations need repeatable ways to connect contributor interest with maintainer needs, and to make that work visible, supported, and durable over time. The Sustaining Open Source Series is one expression of that model. It gives employees a structured path into upstream contribution, gives maintainers concentrated support for work for which they often lack bandwidth, and gives companies a repeatable way to contribute back to help sustain the critical digital infrastructure on which they and others rely. Bloomberg developed this model and has now run this series four times – first with pandas (in partnership with NumFOCUS and NVIDIA), and now with OpenTelemetry (in partnership with the CNCF). The pattern holds true across different project types, languages, and contributor profiles. If your organization is exploring what sustained open source stewardship looks like in practice – not just in theory — we’d welcome that conversation. The outcomes are real, the model is repeatable, and the open source projects that matter most to all of us will be healthier for it. We’ll be sharing full outcomes and learnings at Open Source Summit Europe in Prague this October. To compare notes or explore a collaboration, please reach out to the Bloomberg Open Source Program Office. Bloomberg’s Open Source Program Office runs the Sustaining Open Source Series in partnership with open source foundations and project maintainer communities.