Role: Design authority for the target Fabric platform on a Databricks-to-Fabric migration. Owns the decisions the engineering team builds against and represents iLink in design forums with Microsoft ISD and the client. The role is accountable for the technical decisions that cannot be reversed cheaply — the migration patterns, the parity tolerance model, and the gold-layer design.
Responsibilities:
Target Fabric platform design — OneLake and workspace architecture, capacity model, and the disposition map across the estate
Migration patterns for notebooks, jobs, schemas, tables and the view layer, including the Databricks SQL to Fabric T-SQL dialect conversion
Azure tier decision and design for the regulatory SQL Server estate
Re-platform pattern for the custom applications, including the write-back tier
The parity tolerance model, negotiated and agreed with the client — exact-match classes for regulatory figures, agreed percentages elsewhere
Gold-layer and star schema design suited to Direct Lake
Semantic model architecture across the governed metric estate, with canonical metric definitions agreed with the client
Modernisation sequence and change taxonomy
Technical escalation point; design review and sign-off at each phase gate
Mandatory skills:
Microsoft Fabric at architecture depth — OneLake, lakehouse and warehouse, Direct Lake, capacity and workspace design, deployment pipelines
Databricks at architecture depth — Unity Catalog, Delta Lake internals, Delta Live Tables, Databricks SQL.
Spark SQL semantics across runtimes — type coercion, decimal and timestamp handling, null behaviour.
Medallion architecture and dimensional modelling — Kimball, conformed dimensions, grain decisions, star schemas suited to Direct Lake
Large-scale platform migration — source-to-target mapping, disposition mapping, phased cutover, parity-based acceptance
Client-facing design authority — owning a target architecture, chairing design reviews, holding a technical position under commercial pressure
Azure platform — Entra, Key Vault, networking, private endpoints, Azure SQL Managed Instance