TL;DR: Learn SQL and data modeling first. Then build an ingestion job, transform reliable tables, load a warehouse, test the data, and document lineage and freshness.
Start with SQL and Models
Practice joins, window functions, aggregations, constraints, indexes, and transactions. Model entities and events before choosing a pipeline tool.
Build the Pipeline
Ingest a small event dataset, make the job repeatable, handle duplicates, validate schemas, and separate raw data from cleaned and business-ready tables.
Reliability Is the Skill
Track freshness, row counts, null rates, failures, retries, and schema changes. A dashboard built on silently broken data is worse than no dashboard.
Portfolio Deliverable
Publish the data model, pipeline diagram, sample queries, quality checks, and a short incident runbook. Explain why each transformation exists and who consumes it.