The interview has moved past SELECT. Can you build the table?
Most SQL books teach you to read data. The interviews that decide senior data roles also ask whether you can build it and keep it correct: design a sound table, choose its keys, enforce integrity, change data safely and wrap risky work in a transaction. Table Stakes teaches that build side of SQL from the interviewer's side of the table.
Each chapter opens with the idea in plain language and a diagram. Then, one question per page, it shows what the interviewer is testing, the answer they want to hear, how to say it out loud, a real query with the result it produced, the green and red flags that decide it, and a short "Your turn" task that turns recognition into fluency. Some cells fail on purpose: a CHECK that rejects a negative price, a UNIQUE that blocks a duplicate email. The error message is the lesson.
Inside:
- Tables and data types, keys and relationships, and the constraints that reject bad rows before they reach an analysis
- Inserting and loading data, updating and deleting safely, and transactions and ACID
- Shaping data with CASE and expressions, reshaping tables, and reading the result back
- The rounds: the take-home build task, live design, debugging a constraint error, and SQL injection, stored procedures and dynamic SQL
- A one-page cheat sheet
More than 120 interview questions run on a small commerce schema: customers, products, orders and the order_items table that links orders to products, built with the gaps interviews probe (customers who never ordered, a product never sold, missing emails). Every query was run in the LucidPlexus Workbench on DuckDB, in standard SQL that carries with small changes to PostgreSQL, Snowflake, BigQuery and SQL Server.
Every purchase includes the four tables as CSV files, ready to load.
The Data Interview Series · Interview Edition. LucidPlexus.