Proposal to Adopt Semantic Operator as a Kubeflow Subproject #1023

Proposal

We propose donating and adopting Semantic Operator as a Kubeflow subproject.

Semantic Operator is a Kubernetes operator and stateless semantic server that runs an Apache Ossie semantic layer on existing data platforms. It allows teams to define certified metrics, dimensions, relationships, and governance policies as Kubernetes resources.

AI agents select certified business concepts instead of generating SQL. Semantic Operator validates each request, applies governance, and deterministically generates one SQL statement for StarRocks or Trino.

Why this matters to Kubeflow

AI agents can generate SQL that executes successfully while returning an incorrect business answer. They may select the wrong columns, aggregations, joins, or filters.

Semantic Operator provides a governed data-access layer for agents, applications, BI tools, and Kubeflow workloads. Potential integrations include:

• Kubeflow Pipelines
• KServe
• Kubeflow Model Registry
• Kubeflow Trainer
• Kubeflow Notebooks
• Kubeflow Profiles
• Kubeflow Spark Operator

Donation and governance

We believe Kubeflow can provide vendor-neutral governance and help build a broader community around the project, similar to the governance provided for Kubeflow Spark Operator and Spark History Server MCP.

The proposed destination is:

kubeflow/semantic-operator

The current maintainers, Vara Bonthu (@vara-bonthu) and Manabu McCloskey (@nabuskey), commit to continuing maintenance, supporting the transition, onboarding contributors, and working toward broader community ownership.

Proposal document

Semantic Operator Kubeflow Subproject Proposal

The document is open for comments and community feedback.

Project links

• Repository: https://github.com/KubedAI/semantic-operator
• Documentation: https://kubedai.github.io/semantic-operator
• License: Apache License 2.0

Requested next steps

• Review the proposal and collect community feedback
• Present a project demo during a Kubeflow Community Call
• Identify the appropriate Kubeflow Working Group
• Meet with the Kubeflow Steering Committee
• Submit the corresponding subproject application pull request

Thank you for reviewing this proposal.