Moving From Supply Chain Dashboards to Decision Loops With the Model Context Protocol
A supply chain control tower shows a red tile. A supplier's on-time delivery rate dropped below threshold four days ago. The tile has been…
The archive
Writing, page 7
593 posts on Apache Iceberg, lakehouse architecture, data engineering and applied AI.
Metric Contracts as the Interface AI Agents Actually Need
Two teams present in the same meeting. Sales says pipeline conversion is 24 percent. Finance says it is 19. Both numbers came from the same…
Cross-Cloud Credential Vending in Apache Polaris and the End of Permanent Storage Keys
Pull up the configuration for any Spark cluster that reads a data lake and look for the storage credentials. In most organizations you find…
Designing Policy-Aware Telemetry Tables for AI Systems in Apache Iceberg
Two days ago, on August 2, 2026, the EU AI Act's Article 50 transparency obligations came into application. Many teams had that date…
Defending the Lakehouse Gateway Against Prompt Injection and Data Exfiltration
A support ticket contains a customer's message. Somewhere in that message is a sentence addressed to nobody who works at your company:…
How the Iceberg REST Catalog Turned Into the Lakehouse Control Plane
A dbt run updates a fact table and two dimension tables. The fact table commit succeeds. The second dimension commit fails on a conflict.…
A Migration Playbook for Moving Legacy Warehouses onto Apache Iceberg
The migration plan says twelve weeks. Week fourteen arrives and the team has moved four tables out of six hundred, because table number…
What Zero-Copy Data Sharing Actually Does Between Salesforce, Snowflake, and Databricks
A customer abandons a cart at 8 PM. That event lands in a warehouse table. A marketing platform needs it to trigger a journey, and a…
Apache Polaris 1.7.0 and the Quiet Work of Making a Catalog Trustworthy
A Spark job commits a table update. The catalog writes the change to Postgres. Then the network drops between the catalog and the client,…
Designing Batch Pipelines That Write Well Into Apache Iceberg
The pipeline runs at 2 a.m. It reads yesterday's extract, does its transformations, and writes to an Apache Iceberg table. Six months later…