Data/ML/AIDatabases
Text-to-SQL
Also written as NL2SQL, Natural Language to SQL, Talk to Your Data
Letting people ask questions of a database in plain English, with an AI writing and running the SQL query behind the scenes. It is one of the most common enterprise AI projects and one of the easiest to get subtly wrong.
Think of it like
A data analyst who translates 'how many customers churned last quarter?' into the exact query, except the analyst is a model.
Junior or senior?
Demos are easy; accurate answers on a messy real database are hard.
Senior sounds like
Talks about giving the model table descriptions, testing against known answers, and stopping it from running dangerous or expensive queries.
Ask them
“How did you check the generated queries were returning the right numbers, not just running without errors?”