aws-samples/query-databases-with-natural-language
This project uses the open-source model Mistral Small, deployed in Amazon SageMaker or invoked via API on Amazon Bedrock, to enable users to chat with their database using natural language, without writing any code or SQL query.
Implements a Text2SQL pipeline that generates and executes SQL queries through a two-pass LLM approach: first converting natural language to SQL using the model's schema understanding, then interpreting tabular results back into natural language responses. Supports multiple relational backends (Redshift, RDS, Aurora, Athena, Snowflake) and deploys via SageMaker JumpStart with optional error-recovery retry workflows for failed queries. The orchestration runs in SageMaker Studio notebooks, Lambda, EC2, or ECS with optional Step Functions integration.
No commits in the last 6 months.
Stars
24
Forks
4
Language
Jupyter Notebook
License
MIT-0
Category
Last pushed
Mar 31, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/vector-db/aws-samples/query-databases-with-natural-language"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.