How lequokhanh/text-to-sql-slm works
Project Type and Purpose
• This is a full-stack, microservices-based “text-to-SQL” platform (“text-to-sql-slm”). Users interact via a rich React UI, which calls a Java Spring Boot backend that orchestrates two downstream services:
• slm-embed: manages database connections, schema introspection and metadata.
• slm-engine: runs LLM-driven workflows to translate user text into SQL queries.
• Under the covers, the system also uses persistent storage (MySQL/PostgreSQL), a vector store (Weaviate), and external LLM endpoints for embeddings and completions.
File Structure Highlights
• Top-level services:
• slm-ui (frontend, Vite + React)
• slm-backend (Java/Spring Boot)
• slm-embed (Java/Spring Boot microservice for schema)
• slm-engine (Python FastAPI/Flask service with LLM workflows)
• DevOps config: Dockerfiles, docker-compose.yml (spinning up the four services plus MySQL, Postgres, Weaviate).
• sample_pg_docker: example Postgres DB for local development.