Enabling Data Sovereignty in AI Workflows through Data Mesh Architecture
summary
Claims that a Kubernetes-native Data Mesh delivers data sovereignty for AI by keeping data local while enabling federated queries and model training; uses MinIO for S3 storage, Iceberg for table catalogs, Trino as a federated query engine, Keycloak for identity and Open Policy Agent for policy-as-code, Airflow for orchestration and Jupyter for model work. Demo deploys via a Helm chart, shows role-based masking and access controls working, and outlines federated learning to train local models and aggregate parameters instead of moving raw data.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:06Data sovereignty in AI workflows
- 4:58Centralized data challenges
- 9:19Data Mesh principles
- 16:50Data Mesh for AI and compliance
- 21:48Enforcing sovereignty architecture
- 24:17Demo: storage layer
- 27:06Federated access with Trino
- 30:24Policy enforcement with OPA
- 34:00Data access control demo
- 39:21Data product UI integration
- 40:48Data products as services
- 44:21Q&A
related talks
28mFederated Data Mesh: The Hidden Foundation for Trustworthy AI
34mTransforming Data Processing with Kubernetes: Journey Towards a Self-Serve Data Mesh
43mUntangling Data Mesh: A paradigm shift in Big Data
20mEmbracing Data Mesh: A Modern Strategy for Scalable Data Management
17mData Mesh: The Speedrun
39mKeynote - Data Mesh
48mData Mesh Paradigm Shift in Data Platform Architecture -
36mIntroducing Data Mesh A Paradigm Shift In Analytical Data Management Architecture