All work

MTN Nigeria, Sole Designer & Engineer

SQL Generator

LLM-powered service that converts structured query descriptions into validated SQL for large-scale enterprise datasets.

Enabled product teams to dynamically generate complex queries over large datasets, unlocking flexible data access while preserving safety and validation constraints.

Designed and implemented a natural language-to-SQL generation service for an internal product team requiring flexible access to large-scale datasets.

Enabled dynamic SQL generation based on structured query descriptions, eliminating the need for hardcoded combinations and restrictive data filtering. Implemented schema-aware prompting, post-processing validation, and safety enforcement to ensure reliable and secure query generation.

Developed as part of a rapid AI delivery initiative to support smaller enterprise use cases without full project overhead.

Built with

  • NestJS
  • TypeScript
  • Docker
  • openai-agents