Ten mistakes that make Azure data platforms slow, insecure or unreliable, from tiny files and leaked keys to failure paths that report success, with the fix for each.
Remove secrets from Azure data pipelines: a user-assigned managed identity with least-privilege roles in Bicep, Entra-only SQL access, Key Vault for what's left, Databricks storage credentials, and audit logging.
Enterprise practices for Azure Data Lake Storage Gen2: account topology, redundancy, directory layout, file sizing, RBAC versus ACLs, private endpoints, HNS feature support and lifecycle tiering.
What belongs in the bronze, silver and gold layers of a lakehouse, with a worked PySpark and Delta example, a placement guide for common tasks, and the trade-offs behind how strictly to follow the pattern.
A guided build of a lakehouse-style data platform on Azure: ADLS Gen2, Data Factory, Databricks with Unity Catalog, Key Vault and Azure Monitor, deployed with Bicep and wired together with managed identities.