SQL query solver

Your personal SQL bot

Intro:

'SQL query solver' is a fully localized python-backend system which efficiently executes RDB (relational database) queries
ie: What is the maximum salary? and fetches query results from your local MySQL server.

Problem considered:

Storing big data into databases,generaly Relational databases (MySQL) is an efficient and common approach but, to execute a 'SQL query' , inorder to fetch meaningful information , is a challenging task for non-technical person ,therefore I developed 'SQL query solver' which enables the user to enter a query in Natural Language (ie: what is the name of highest paying employee?) and fetches the result from user's local database. In simple words , NO MORE "SELECT * FROM emp WHERE sal IN(SELECT MAX(sal) FROM dept WHERE....).." 😊🚀

The best part is this setup is fully localalized (offline) which implies that your data is fully secured from unauthorized access or malware attacks, which is a plus for low-cost organizations or startups.

Technical stack:

◾python streamlit --- for web-UI
◾Space efficient local LLM: deepseek-r1:1.5b via Ollama version is 0.5.13
◾sqlalchemy and pymysql --- for MySQL database connectivity
◾python requests --- for conversation with local LLM

Workflow in simple steps:

After connecting to MySQL server:
1. User enters the question in Natural Language on the UI
2. Local LLM (in this case: deepseek r1) converts the question into an executable SQL query
3. SQL query is executed on MySQL server
4. Result is fetched and displayed back to the user

Modifications as per requirement:

◾It is suggested to integrate a local LLM having more parameterized functions (ie: >1.5 billion parameters), to ensure efficiency in complex SQL queries and less errors.

Steps for installation:

1. Install python dependancies: (~200 MB)

        CMD:  pip install streamlit pandas pymysql sqlalchemy requests

2. Install Ollama:

   Download from: https://ollama.ai/download Run the .exe installer

3. Download AI model:

        CMD: ollama pull deepseek-r1:1.5b
   

4. Set up MySQL:

   Open MySQL Workbench

   Run:

        SQL:  ALTER USER 'root'@'localhost' IDENTIFIED WITH mysql_native_password BY 'your_password';
              FLUSH PRIVILEGES;
   

5. Clone github repo:

   Copy-paste the code file to local drive

        CMD: streamlit run app.py