Free Microsoft Fabric Tutorial Series

Every Fabric skill you build later depends on getting the architecture right first. This module covers why OneLake is the platform’s central pillar, how it differs from a traditional ADLS Gen2 data lake, and the “one copy” principle that lets every workload read the same underlying data without duplicating it.

Data engineering is where raw data becomes something usable. This module covers ingestion and transformation from both angles — the low-code path with Dataflow Gen2, and the code-first path with Spark notebooks — plus Data Mirroring, which replicates operational databases into OneLake without writing ETL code at all.

For SQL-first teams, the Warehouse is often the most familiar environment in Fabric. This module covers T-SQL development on the serverless Polaris engine, the Medallion architecture pattern for structuring Bronze/Silver/Gold layers, and the open table-format question — Delta Lake vs. Apache Iceberg — that increasingly comes up in platform decisions.

Modern analytics isn’t limited to batch processing. This module covers Fabric’s streaming stack end to end: Eventstreams for ingestion and routing, KQL databases for sub-second time-series querying, the Real-Time Hub as the tenant-wide catalog of streaming data, and Data Activator for turning detected patterns into automated action.

Data eventually has to be served to people. This module covers Power BI’s semantic modeling layer in depth — DAX beyond the basics, version-controllable model definitions with TMDL, and Direct Lake mode, which is powerful but requires specific optimization to avoid a silent fallback to slower query performance.

Security in Fabric operates in layers — workspace permissions, OneLake data access roles, and semantic-model RLS all interact, and getting the interaction wrong is how data gets exposed or performance silently degrades. This module also covers the two OneLake security failure modes we’ve documented in detail: an exposure risk and a permanent Direct Lake performance conflict.

This module covers Fabric’s generative AI layer as it actually applies to day-to-day work — grounding AI answers in your own OneLake data with Data Agents and RAG patterns, preparing semantic models so Copilot answers questions accurately instead of guessing, and writing effective prompts for the AI-assisted tools now built into the platform.

Fabric’s newest surface area extends beyond Copilot into a broader agent platform — domain-specific “IQ” layers, native apps, and the protocols that let external tools connect to Fabric data directly. This area moves fast; treat this module as a map of what exists today rather than a finished, settled feature set.

Once you’ve worked through the modules relevant to your role, these guides turn that knowledge into a certification and interview-ready form. If you’re not sure which certification to target, start with the comparison guide before picking an exam.