Machine Learning
QueryBank — Banking Text-to-SQL Chatbot
Natural-language chatbot that converts banking questions into SQL plus auto-selected charts.
- TypeScript · Python
Overview
A text-to-SQL chatbot for a banking PostgreSQL database. Google Gemini reads schema documentation and emits both an SQL query and a matching chart configuration; results render as tables and auto-selected bar, line, pie, or scatter visualizations across customer, loan, and analytics views. Combines a Python query-generation core with a TypeScript/Next.js dashboard exposing landing, login, chatbot, SQL, and analytics pages.
Key highlights
- Schema-aware text-to-SQL generation backed by Google Gemini
- Automatic chart-type selection (bar / line / pie / scatter) per query result
- PostgreSQL backend with PL/pgSQL stored procedures for analytics views
- Next.js dashboard with login, chat, SQL workspace, and analytics surfaces
- Domain-specific guardrails for banking customer and loan datasets
Tech stack
- Next.js
- PostgreSQL
- Google Gemini
- Python
- PL/pgSQL
Topics
- #text-to-sql
- #llm
- #gemini
- #chatbot
- #data-visualization
- #banking
- #analytics