This is a local text-to-SQL converter built using Streamlit and Ollama, powered by Llama3.2. It enables you to interact with your database schema using natural language prompts, generating accurate SQL queries in return.
✅ Key Benefits:
- Local Execution: No need to share proprietary or sensitive data with external LLMs.
- Database-Aware: Pre-configured for BigQuery and Redshift schemas, following their specific SQL dialects.
- Customizable: Easily extend the schema and model selection to match your environment.
💡 This tool is particularly useful when working with confidential datasets or offline environments, where external LLM access is restricted.
Screenshots:
Streamlit App
Download and install Ollama to run local LLMs:
Ollama Installation
Make sure you have installed minstral and ollama3.2 libraries
(see ollama documentation)
git clone https://github.com/Ethanlchandler/local_textqlOn Linux/macOS
python3 -m venv venv
source venv/bin/activatepip install -r requirements.txtstreamlit run main.pyThis tool can be extended or forked to suit various data engineering and analytical needs:
-
Data Exploration:
Ask complex SQL questions in natural language to explore datasets without manually writing SQL. -
ETL Pipeline Testing:
Quickly generate SQL queries for ETL validation by describing the logic in natural language. -
SQL Query Learning:
Use it as an educational tool to learn SQL by converting plain questions into SQL syntax. -
Custom Database Support:
Add support for more SQL dialects (PostgreSQL, Snowflake, etc.) by extending the schema context. -
Integration with CI/CD:
Automate query generation as part of CI/CD pipelines to validate data changes.
- Language Model: Llama3.2 (or Mistral) via Ollama
- Framework: Streamlit
- Languages: Python
- Databases: BigQuery and Redshift
Ethan Chandler
✅ Let me know if you want further modifications, additional sections, or more details!