Frustrating SQL tutorials — teach applied SQL with realistic bank scenarios

Many SQL learners and hiring teams complain that existing tutorials use toy datasets and contrived exercises that do not reflect messy, regulated production environments; this is especially true for 12 million developers and data professionals (and growing numbers of PMs and analysts) who collectively spend roughly $500 per year on training, yielding a $6.0B market. The result is practitioners who can write SELECTs on clean tables but struggle with real banking problems like multi-entity reconciliations, ledger rollups, fraud scoring across streaming partitions, and compliance-driven data redaction. The product would be an interactive learning and assessment platform centered on realistic bank scenarios: canonical messy schemas, time-series and partitioned workloads, audit trails, and role-based exercises (analyst, investigator, engineer) with automated grading that uses rule-based checks plus LLM-assisted feedback and variations generation. Features would include sandboxed SQL execution with explain-plan feedback, employer-aligned assessments and badges, scenario mutation for unlimited practice, and optional integrations with applicant screening tools. This is an attractive moment: skills-based hiring is increasing demand for practical SQL assessments, generative AI materially lowers the marginal cost of creating varied, domain-specific exercises and grading, and the rise of non-engineer data roles expands the TAM. Market indicators are favorable (market score 92/100, revenue potential 80/100), but the business faces real challenges around legal/regulatory sensitivity when modeling financial scenarios, the engineering cost of secure, scalable sandboxes, and the need to prove employer signal over generalist platforms.