I did not start in IT. I started on the business side, and I moved into data engineering without a degree, which meant every door I walked through, I had to interview my way through. I am finishing that degree now, years after it stopped being the thing standing in my way.

So I know what the process actually feels like. The screening call. The live SQL round with someone watching your cursor blink. The question you answer confidently and get wrong, and the silence afterward. There is a real gap between knowing SQL and proving it in forty-five minutes to a stranger who has already decided to be skeptical, and nobody warns you about that gap until you are standing in it.

Five years later I own a data platform: a dbt Medallion architecture on BigQuery and the change data capture pipelines that feed it, with SQL Server transaction logs streamed through Kafka into the warehouse in near real time. I built and own a natural language analytics application that generates SQL against that warehouse, using retrieval over schema and business context, vector search, and MCP servers. I also own the data infrastructure behind a GenAI team.

Along the way: a SQL Server estate migrated to BigQuery covering roughly 200 SSIS packages and 1,300 reports across billions of records, an address ingestion pipeline rebuilt in Apache Beam and taken from 38 hours to under one, BigQuery costs cut by about 70 percent, and a call analytics dashboard serving thousands of users under row level security.

I teach the SQL that decides interviews, because I learned it the expensive way. Not trivia, not syntax drills. The join that silently doubles your revenue. The NULL that makes a correct-looking query return nothing at all. The query that gives you a plausible number instead of an error, which is the one that actually costs you the offer. If you are trying to get in without the conventional background, this is the material I wish someone had handed me.