mris.dev

What Is It?

Snowy is a Streamlit-based conversational Business Intelligence tool that interprets natural language queries, translates them to SQL, and visualizes data from Snowflake. It uses Google’s Gemini Pro LLM to power the natural language to SQL translation.

Why Build It?

Business intelligence tools typically require users to know SQL or navigate complex UIs. Snowy lets anyone ask questions in plain language and get visualized answers — democratizing data access across the organization.

How It Works

User asks a question in natural language
LLM translates the query to Snowflake SQL
Query executes against Snowflake with secure key-pair authentication
Results visualized with adaptive chart types based on the data
Query transparency — users see the SQL that was generated

Features

Natural language interface — Ask questions in plain language
LLM-powered translation — Uses Google Gemini Pro for SQL generation
Secure Snowflake connectivity — Key-pair authentication, no passwords
Adaptive visualization — Chart types adapt to query results
Query transparency — See the SQL behind every answer
Error handling — Graceful handling of failed queries

Tech Stack

Layer	Technology
Language	Python 3.8+
UI	Streamlit
Database	Snowflake
LLM	Google Gemini Pro
Auth	Key-pair authentication

Prerequisites

Python 3.8+
Snowflake account with appropriate permissions
Google Cloud API key for Gemini Pro
Private key for Snowflake authentication

Quick Start

Installation

git clone gitlab.sicepat.tech/ilmimris/snowy.git
cd snowy
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Running

streamlit run app.py

Use Cases

Self-Service Analytics — Non-technical users can query data
Executive Dashboards — Natural language queries for business metrics
Data Exploration — Quick ad-hoc analysis without writing SQL
Reporting — Generate reports from natural language descriptions
Data Validation — Verify data quality with simple questions

Example Queries

“What were total sales last month?”
“Show me revenue by region for Q1”
“Which products had the highest return rate?”
“Compare customer acquisition costs by channel”

Snowy