Introduction
My initial goal was to learn how to create a Copilot assistant that could support asking questions on Snowflake. Since I only had a little knowledge of Snowflake, I decided to take this as an opportunity to build a learning journey. Using GitHub Copilot, I generated a series of labs to guide the process. I followed these labs step by step, worked through the exercises, and asked deeper questions along the way to dive further into the topic.
This project https://github.com/phaledang/learn-snowflake started with this prompt, using GitHub Copilot:
“I want to create a lab project with README files and presentation slides for step-by-step learning of Snowflake. The labs will be numbered (lab01, lab02, …) with a sample for each day of learning. I will use SQL and Python with Snowflake. The project should include instructions on how to start a trial and how to use Snowflake.”
From that one prompt, Copilot scaffolded a complete workspace—numbered lab folders, each with a README, sample SQL, Python snippets. Over time, the structure matured through real-world iteration: adding reproducible environments, refining queries, and layering richer examples.
The labs are designed like building blocks. Each module introduces one concept and feeds into the next, creating a smooth learning curve.
Challenges and Fixes Along the Way
These labs aren’t about theory—they’re all about hands-on practice. I signed up for a free 30-day trial on Snowflake Trial, worked through each step, created guidelines along the way, and updated the code samples Copilot generated to make it work. To make things easier to follow, step-by-step instructions and screenshots are added.
FastAPI port conflicts → switched from 8000 to 8080.
CORS failures → explicit CORSMiddleware added for browser apps.
Annoying pandas warnings → refactored to SQLAlchemy for clean reads.
Connection chaos → unified snowflake_connection.py utility.
Dependency mess → per-lab venv/ with pinned requirements.
Debugging → VS Code launch configs for some labs.
Supporting scripts like debug_connection.py and check_table.py were introduced to help diagnose problems.
Some Copilot prompts that shaped the repo
Create a syntax reference md file for every Snowflake SQL script used in this folder.
Change FastAPI server to use port 8080 instead of 8000.
Add CORS handling so React frontend requests stop failing.
Refactor Snowflake connector code to use SQLAlchemy and remove pandas warnings.
Support SNOWFLAKE_CONNECTION_STRING in .env
Can’t load plugin: sqlalchemy.dialects:snowflakeinclude lab05 debug into lauch.json
Include lab05 debug into lauch.json
Output
The README files, syntax references, and code samples in the labs provided valuable support throughout my Snowflake learning journey. They offered structured guidance across key areas, including:
- Snowflake architecture and core concepts
- SQL operations and advanced functions
- Data loading and warehouse management
- Python integration with Snowflake
- Data sharing and security features
- Performance optimization techniques
- Real-world data engineering projects
While some code samples required adjustments to run successfully, the process of troubleshooting and resolving these issues significantly deepened my understanding of Snowflake and its practical applications.
Lab-07 – Chat with Snowflake data – Output
