Proposal to start the Agents WG
Chase Christensen
Hello Kubeflow community!
I am proposing we start an Agentic Workflows Working Group, as discussed [here].
This working group will focus on the deployment, maintenance, and operationalization of autonomous agentic workflows within Kubeflow. While the wg-ml-experience governs direct-to-IDE developer integrations, this WG will focus specifically on the autonomous agent experience—building the infrastructure and tooling that allows agents to independently execute operational tasks (e.g., "investigate this Spark app") and serve integrated solutions within Kubeflow.
With projects like the docs-agent, Spark MCP, and more already in motion, I believe it is time we start thinking strategically about how we can ensure our ecosystem is fully agent-friendly. This means helping users fine-tune better instruct models, utilize agents seamlessly with our components, and run agents securely alongside Kubeflow.
The world is at our fingertips, and with graduation coming up, the iron is hot! Please leave a comment, suggest edits to the charter, and help us build the future of an agentic Kubeflow.