A realistic six-week plan at roughly 8 hours per week. Keep a Databricks workspace open (free Community Edition or a trial) and do every task hands-on. Add two weeks if you are new to Databricks.
| Week | Focus | Checkpoint |
|---|---|---|
| 1 | Platform basics: workspace, clusters/compute, notebooks, Databricks SQL, the medallion design | You can run code on a cluster and explain bronze/silver/gold |
| 2 | Delta Lake: managed vs external tables, reads/writes, time travel | You can create and query Delta tables and recover an earlier version |
| 3 | Ingestion with Auto Loader; transformations with Spark SQL and PySpark | You can incrementally load files and clean them into a silver table |
| 4 | Structured Streaming basics; aggregate into gold tables | You can explain batch vs streaming and build a gold aggregate |
| 5 | Lakeflow Declarative Pipelines (DLT) with expectations; Jobs/Workflows orchestration | You can build a pipeline with a quality check and schedule it |
| 6 | Unity Catalog governance; full-length timed reviews across all five areas | You consistently pass timed reviews on fresh questions |
Final tips
Pipelines and streaming (Weeks 3–5) are the part most people underestimate - give them the most hands-on time, and build one end-to-end pipeline yourself rather than only reading. Know the split between Lakeflow Declarative Pipelines (which declare and maintain tables) and Jobs/Workflows (which schedule and orchestrate). Governance with Unity Catalog gives easy marks once you have set up a catalog and a permission yourself. Remember this exam is platform-specific: if your employers use Snowflake or BigQuery, a platform-specific certification is the better target. Avoid “exam dump” sites - they breach Databricks policy. And do not trust third-party claims of an exact passing score, per-section weights, or an exam code; Databricks does not publish them.