Federated Data Mesh: The Hidden Foundation for Trustworthy AI
summary
Claim: a federated Data Mesh built on Google Cloud (Databricks as logical layer, BigQuery + GCS for storage, Terraform and CI/CD for ops) keeps infrastructure centralized while pushing data ownership, governance and metadata to domains. Implementation runs two years at Walter Group, enforces metadata, privacy tags and log monitoring, produces domain-specific lakes and data products, and cuts the central ticket backlog by about 50% backlog reduction, aiming to deliver more trustworthy AI by removing data ambiguity.
AI-written summary · about
chapters (13)
- 0:00Federated Data Mesh for AI
- 1:05Speaker background and Walter Group overview
- 2:33Trustworthy AI beyond hype
- 3:58Data mesh challenges
- 6:03Federated data mesh solution
- 9:19Enabling trustworthy AI
- 12:27Walter Group example
- 18:16AI project success stats
- 19:19Implementation key points
- 21:50Q&A
- 23:35Q&A
- 25:43Q&A
- 27:30Closing
related talks
43mUntangling Data Mesh: A paradigm shift in Big Data
45mEnabling Data Sovereignty in AI Workflows through Data Mesh Architecture
39mData Mesh: Why, What and How
48mData Mesh Paradigm Shift in Data Platform Architecture -
47mData Mesh - a hot air balloon?
20mEmbracing Data Mesh: A Modern Strategy for Scalable Data Management
01h 00mData Mesh: Beyond the Data Lake
38mData Mesh, Now