Natural-language SQL and analytics playground with optional provider generation. Explore deterministic sample schemas, inspect SQL, and run supported queries locally in SQLite.
Explore synthetic datasets with reviewed query templates, local SQLite execution, and visible validation
Browser-local intent matching separates supported, ambiguous, unsupported, and exploratory requests
Typed requests plus a narrow, tested read-only SQLite policy
Table grains, columns, keys, and relationship cardinality remain inspectable
TypeScript keeps clarification, supported answers, and exploration distinct
Separate generation, review, and explicit local execution states
Synthetic fixtures and browser policy are scoped honestly; the provider backend is optional
Optional typed generation API; provider output is returned as not executed
Deterministic synthetic schemas with explicit, read-only local execution
Generation source, schema, execution status, and truncation stay visible
Choose a sample schema, inspect curated or generated SQL, then run it explicitly against local fixture data.
This portfolio project demonstrates analytics and applied-AI engineering. Curated examples need no key; optional live generation requires configured backend credentials. Accepted SQL executes locally only against bundled synthetic sample data. No aggregate usage or model-accuracy metrics are claimed. View the full implementation at github.com/cbratkovics/sql-genius-ai