In this episode of The Landscape, we spoke with Valentina Rodriguez Sosa, Principal Architect at Red Hat and an active contributor to the Kubeflow project. Valentina is passionate about simplifying machine learning workflows and fostering a collaborative, open-source community. Kubeflow is an incubating CNCF landscape project.
Valentina is excited to showcase how Kubeflow helps organizations streamline their AI/ML pipelines on Kubernetes. As a release manager for Kubeflow, she highlighted the platform’s ability to integrate popular tools like TensorFlow and PyTorch while providing powerful features like a model registry for version control and auditability. Valentina is particularly proud of Kubeflow’s role in making AI accessible to everyone, from newcomers to experts.
For those new to Kubeflow, Valentina recommends exploring its community resources, including Slack channels and GitHub repositories, as well as attending the public meetings. “Contributing to Kubeflow is an excellent opportunity to grow your skills, collaborate with passionate people, and make an impact in the AI space,” she shared.
This episode is sponsored by OVHcloud.
Valentina Rodriguez Sosa:
Hi, I’m Valentina Rodriguez Sosa, a Principal Architect at Red Hat. I focus on OpenShift but also contribute to CNCF projects. I’m involved in the Kubeflow project and serve as part of the release management team.
Bart:
That’s great! What’s your backstory? How did you get involved in Kubeflow?
Valentina:
With everything AI-related being so prominent, I wanted to give back to the community and spend more time in this space. I started exploring projects to contribute to and saw that Kubeflow was looking for release management help. To do that, I needed to begin contributing, so I started with project documentation. Being detail-oriented, I focused on fixing issues and creating pull requests.
The community was fantastic—they welcomed me with open arms. My contributions were reviewed quickly, and the team supported me every step of the way. After some time, I began shadowing Ricardo, the release manager, and eventually started leading meetings. The release management team, along with the broader community, has been incredibly open and diligent in keeping everything updated.
We even have a public dashboard showing all the features being worked on and their progress. Community meetings are open to everyone, making it easy to get involved.
Bart:
For those unfamiliar, what problems does Kubeflow solve?
Valentina:
Kubeflow simplifies AI and ML workflows, making them accessible for everyone—from newcomers to experts. It provides an end-to-end platform where you can build, train, and deploy models seamlessly. Kubeflow handles complexities like tool selection, versioning, and production deployment, so users can focus on building their models.
One feature I love is the model registry. It provides a centralized place to track model versions, metadata, and audit trails. This ensures that teams are working with verifiable and reliable models.
Bart:
Are there any updates or new features in the works?
Valentina:
We’re working on the 1.10 release. While I can’t share specific details yet, I can say the team is focused on updating dependencies, improving security, enhancing documentation, and refining the user experience—especially for the model registry and integrations with accelerators.
Bart:
How can people get involved in Kubeflow?
Valentina:
It’s really easy to start. Visit the Kubeflow website and check the “About” section for community resources. Join our Slack channel, attend community meetings, or look for beginner-friendly issues on GitHub.
We’re always looking for contributors—whether you’re technical or not. Documentation updates, for example, are a great way to start. Contributing to Kubeflow is a fantastic opportunity to learn new skills, explore different roles, and grow your career.
Bart:
Thank you so much for sharing your journey and insights, Valentina!
Valentina:
Thank you! Looking forward to welcoming more contributors to the Kubeflow community.