My favorite ways to learn Snowflake hands-on (mostly free): a starter guide
I've been working with Snowflake since late 2025, run our data platform at an insurtech, and help out in the Snowflake community. People often ask "where do I start?", so here's my personal shortlist, in the order I'd suggest going through it. All of it is hands-on or at least very practical.
1. Hands-On Essentials workshops (free, self-paced)
Start here if you're new to Snowflake or to data work in general. These are online, self-paced workshops with short animated scenarios, real labs in a trial account, and knowledge checks. Your lab work is auto-graded, so you get immediate feedback on whether you actually did it right.
No prerequisites, and it costs nothing apart from a free trial account. Overview: snowflake.com
2. Snowflake University / Learn (learn.snowflake.com)
This is the official catalog. It has learning tracks (including SnowPro Core prep), self-paced courses and instructor-led training (ILT).
- Free on-demand: the "Level Up" series (concepts, architecture, data loading, resource monitoring, Snowpark, Native Apps), FinOps for Snowflake, Cost Governance, Gen AI courses, SnowConvert, and Spark → Snowpark migration.
- Paid: Snowflake Foundations on-demand, the SnowPro Core Certification Prep Course, and most ILT courses (Fundamentals 4-day, Administering Snowflake, Data Engineer, Architect, Data Analyst, Snowpark, Native Apps, Performance Tuning and more).
My tip: do the free stuff first, then decide if a paid course is worth it for your role. Catalog: learn.snowflake.com
New: some courses are now offered in German too (e.g. learn.snowflake.com). Tech people work in English anyway, but I expect this to matter once business users and non-engineers start working with Snowflake's AI features.
3. "Snowflake on Snowflake" webinars
Recorded sessions where Snowflake's own teams (IT, Finance, Sales, Marketing, Security, Product) show how they use Snowflake internally. It's less "how do I write a COPY INTO" and more "what does this look like in a real company". Examples:
- Saving ~$5M in SaaS spend with AI on Snowflake
- Lead scoring with Snowpark, Cortex and the Model Registry
- Cloud finance management and cost attribution
- Automating SOX / internal controls
- Using Snowflake as a security data lake
- Account 360 / data enrichment in Salesforce
You can filter by department, product area and region. Great for getting ideas for your own use cases: snowflake.com
4. DataCamp's Snowflake courses
If you prefer interactive in-browser exercises, DataCamp has a solid Snowflake catalog (built together with Snowflake):
- Beginner: Introduction to Snowflake (2h), Snowflake Architecture (3h), Data Manipulation in Snowflake, Data Pipeline Automation (COPY INTO, Snowpipe, streams, tasks), Management, Governance & Collaboration
- Intermediate: Snowflake SQL, Data Modeling, Window Functions, Modern Data Engineering, Apache Iceberg, Generative AI in Snowflake, Building AI Agents
- Tracks: Snowflake Foundations (~7h), SnowPro Core exam prep (~9h), Associate Data Engineer (~32h), Snowflake AI Fundamentals (~6h)
- Newer content also covers Cortex Code (CoCo)
You do need basic SQL (sorting, grouping, aggregation, joins) before you start. Their Snowflake hub: datacamp.com
5. Build a Snowflake Docs agent in your own account (about 5 minutes)
My favorite "learn by doing" exercise: build an agent that answers questions about Snowflake from the official documentation, with citations. You only need three building blocks:
- Snowflake Documentation CKE: a free Cortex Knowledge Extension, i.e. a shared Cortex Search Service, that you install from the Snowflake Marketplace.
- Cortex Agents: configured in Snowsight under AI & ML » Agents.
- Snowflake Intelligence / Snowflake CoWork: the chat UI where you talk to your agent.
Steps:
1. Install the listing. Get "Snowflake Documentation" from the Marketplace and grant imported privileges on the database to the roles that should use it.
2. Create the agent. Create an agent (e.g. snowflake_docs) in any database and schema. Add a Cortex Search tool that points at CKE_SNOWFLAKE_DOCS_SERVICE, set the ID column to SOURCE_URL for clickable citations and the Title column to DOCUMENT_TITLE. Add short column descriptions to improve retrieval.
3. Write the instructions. In the orchestration instructions, tell the agent to answer only when it can cite a reference.
4. Make it visible. If your account uses a Snowflake Intelligence object (check with SHOW SNOWFLAKE INTELLIGENCES;), add the agent with ALTER SNOWFLAKE INTELLIGENCE SNOWFLAKE_INTELLIGENCE_OBJECT_DEFAULT ADD AGENT <db.schema.agent>;. If not, users automatically see all agents they have access to.
5. Try it. Open Snowflake Intelligence, pick your agent and ask: "How do I create an agent?"
In one exercise you touch Marketplace, Cortex Search, Agents and RBAC, and you end up with a docs assistant for everything else you learn.
Fun fact: I couldn't find a proper Snowflake doc page for this exact setup, so I asked my own Docs agent. It happily double-checked its own build instructions, so yes, the agent verified the guide for its own creation. Peak recursion.
Note: Older guides say agents must live in snowflake_intelligence.agents. Snowflake has deprecated that schema for controlling which agents users see, so use the Snowflake Intelligence object instead.
Sources:
- Original walkthrough by Chris Cardillo: medium.com
- Cortex Knowledge Extensions: docs.snowflake.com
- Configure agents and add a Cortex Search tool: docs.snowflake.com
- Deploy agents and control visibility in Snowflake Intelligence: docs.snowflake.com
6. Snowflake Guides / Quickstarts (hands-on tutorials)
Quickstarts are interactive, step-by-step tutorials and self-serve demos that you follow in your own account. They've since been merged with the old "Solutions Center" into what Snowflake now calls Snowflake Guides, published at snowflake.com. A typical guide walks you through Overview, Prerequisites, What You Will Build, What You Will Learn, What You Learned, Conclusion and Resources, so you know upfront what you need and what you'll end up with.
Things I like about them:
- Open source: every guide is a markdown file in the public GitHub repo Snowflake-Labs/sfquickstarts (Apache-2.0 license, under site/sfguides/src/), so you can read the source, follow along, or browse the whole collection at once.
- Fix it yourself: found a typo or an outdated step? You can open a pull request. Guides are not covered by Snowflake Support, so PRs are the way to report errors.
- Contribute your own: fork the repo, copy the markdown template, add your guide and open a PR. Automated checks validate the id, categories and language tag, and the DevRel team reviews it. This is a nice entry point into the community if you've built something worth sharing.
- Low friction for learners: guides can be used anonymously and your progress is saved automatically.
If you're into ML, the Snowflake docs also have a curated "Getting Started" page for Snowflake ML, with 15 guides sorted into end-to-end examples, model development and MLOps (nine of them are Beginner level), for example Build an End-to-End ML Model in Snowflake, Getting Started with Snowflake Notebooks on Container Runtime or Introduction to Snowflake Feature Store: docs.snowflake.com
How I'd combine it
1. Start with the Hands-On Essentials workshops (Data Warehousing first) to get the basics.
2. Build the Docs agent and use it as your study buddy.
3. Follow a few Quickstarts / Guides for the topic you care about most (Snowpark, Notebooks, ML, Cortex...). They're the fastest way to build something real.
4. Watch a few "Snowflake on Snowflake" sessions for real-world context.
5. Use DataCamp or the Level Up courses to go deeper where you need it (SQL, pipelines, Snowpark, AI).
6. If you want a certification, look at the SnowPro Core prep material.
Also: Snowflake Expedition (free virtual event) runs November 3–6, with technical tracks if you want live content.
What did I miss? Which resources helped you most? Happy to hear your favorites, especially for Cortex / AI topics.
P.S. If you want to work towards badges in a structured way, these two landing pages are a good starting point:
- achieve.snowflake.com
- developerbadges.snowflake.com
34 points · 5 comments
Mentions
My favorite ways to learn Snowflake hands-on (mostly free): a starter guide I've been working with Snowflake since late 2025, run our data platform at an insurtech, and help out in the Snowflake community. People often ask "where do I start?", so here's my personal shortlist, in the order I'd suggest going through it. All of it is hands-on or at least very practical. 1. Hands-On Essentials workshops (free, self-paced) Start here if you're new to Snowflake or to data work in general. These are online, self-paced workshops with short animated scenarios, real labs in a trial account, and knowledge checks. Your lab work is auto-graded, so you get immediate feedback on whether you actually did it right. No prerequisites, and it costs nothing apart from a free trial account. Overview: snowflake.com 2. Snowflake University / Learn (learn.snowflake.com) This is the official catalog. It has learning tracks (including SnowPro Core prep), self-paced courses and instructor-led training (ILT). - Free on-demand: the "Level Up" series (concepts, architecture, data loading, resource monitoring, Snowpark, Native Apps), FinOps for Snowflake, Cost Governance, Gen AI courses, SnowConvert, and Spark → Snowpark migration. - Paid: Snowflake Foundations on-demand, the SnowPro Core Certification Prep Course, and most ILT courses (Fundamentals 4-day, Administering Snowflake, Data Engineer, Architect, Data Analyst, Snowpark, Native Apps, Performance Tuning and more). My tip: do the free stuff first, then decide if a paid course is worth it for your role. Catalog: learn.snowflake.com New: some courses are now offered in German too (e.g. learn.snowflake.com). Tech people work in English anyway, but I expect this to matter once business users and non-engineers start working with Snowflake's AI features. 3. "Snowflake on Snowflake" webinars Recorded sessions where Snowflake's own teams (IT, Finance, Sales, Marketing, Security, Product) show how they use Snowflake internally. It's less "how do I write a COPY INTO" and more "what does this look like in a real company". Examples: - Saving ~$5M in SaaS spend with AI on Snowflake - Lead scoring with Snowpark, Cortex and the Model Registry - Cloud finance management and cost attribution - Automating SOX / internal controls - Using Snowflake as a security data lake - Account 360 / data enrichment in Salesforce You can filter by department, product area and region. Great for getting ideas for your own use cases: snowflake.com 4. DataCamp's Snowflake courses If you prefer interactive in-browser exercises, DataCamp has a solid Snowflake catalog (built together with Snowflake): - Beginner: Introduction to Snowflake (2h), Snowflake Architecture (3h), Data Manipulation in Snowflake, Data Pipeline Automation (COPY INTO, Snowpipe, streams, tasks), Management, Governance & Collaboration - Intermediate: Snowflake SQL, Data Modeling, Window Functions, Modern Data Engineering, Apache Iceberg, Generative AI in Snowflake, Building AI Agents - Tracks: Snowflake Foundations (~7h), SnowPro Core exam prep (~9h), Associate Data Engineer (~32h), Snowflake AI Fundamentals (~6h) - Newer content also covers Cortex Code (CoCo) You do need basic SQL (sorting, grouping, aggregation, joins) before you start. Their Snowflake hub: datacamp.com 5. Build a Snowflake Docs agent in your own account (about 5 minutes) My favorite "learn by doing" exercise: build an agent that answers questions about Snowflake from the official documentation, with citations. You only need three building blocks: - Snowflake Documentation CKE: a free Cortex Knowledge Extension, i.e. a shared Cortex Search Service, that you install from the Snowflake Marketplace. - Cortex Agents: configured in Snowsight under AI & ML » Agents. - Snowflake Intelligence / Snowflake CoWork: the chat UI where you talk to your agent. Steps: 1. Install the listing. Get "Snowflake Documentation" from the Marketplace and grant imported privileges on the database to the roles that should use it. 2. Create the agent. Create an agent (e.g. snowflake_docs) in any database and schema. Add a Cortex Search tool that points at CKE_SNOWFLAKE_DOCS_SERVICE, set the ID column to SOURCE_URL for clickable citations and the Title column to DOCUMENT_TITLE. Add short column descriptions to improve retrieval. 3. Write the instructions. In the orchestration instructions, tell the agent to answer only when it can cite a reference. 4. Make it visible. If your account uses a Snowflake Intelligence object (check with SHOW SNOWFLAKE INTELLIGENCES;), add the agent with ALTER SNOWFLAKE INTELLIGENCE SNOWFLAKE_INTELLIGENCE_OBJECT_DEFAULT ADD AGENT <db.schema.agent>;. If not, users automatically see all agents they have access to. 5. Try it. Open Snowflake Intelligence, pick your agent and ask: "How do I create an agent?" In one exercise you touch Marketplace, Cortex Search, Agents and RBAC, and you end up with a docs assistant for everything else you learn. Fun fact: I couldn't find a proper Snowflake doc page for this exact setup, so I asked my own Docs agent. It happily double-checked its own build instructions, so yes, the agent verified the guide for its own creation. Peak recursion. Note: Older guides say agents must live in snowflake_intelligence.agents. Snowflake has deprecated that schema for controlling which agents users see, so use the Snowflake Intelligence object instead. Sources: - Original walkthrough by Chris Cardillo: medium.com - Cortex Knowledge Extensions: docs.snowflake.com - Configure agents and add a Cortex Search tool: docs.snowflake.com - Deploy agents and control visibility in Snowflake Intelligence: docs.snowflake.com 6. Snowflake Guides / Quickstarts (hands-on tutorials) Quickstarts are interactive, step-by-step tutorials and self-serve demos that you follow in your own account. They've since been merged with the old "Solutions Center" into what Snowflake now calls Snowflake Guides, published at snowflake.com. A typical guide walks you through Overview, Prerequisites, What You Will Build, What You Will Learn, What You Learned, Conclusion and Resources, so you know upfront what you need and what you'll end up with. Things I like about them: - Open source: every guide is a markdown file in the public GitHub repo Snowflake-Labs/sfquickstarts (Apache-2.0 license, under site/sfguides/src/), so you can read the source, follow along, or browse the whole collection at once. - Fix it yourself: found a typo or an outdated step? You can open a pull request. Guides are not covered by Snowflake Support, so PRs are the way to report errors. - Contribute your own: fork the repo, copy the markdown template, add your guide and open a PR. Automated checks validate the id, categories and language tag, and the DevRel team reviews it. This is a nice entry point into the community if you've built something worth sharing. - Low friction for learners: guides can be used anonymously and your progress is saved automatically. If you're into ML, the Snowflake docs also have a curated "Getting Started" page for Snowflake ML, with 15 guides sorted into end-to-end examples, model development and MLOps (nine of them are Beginner level), for example Build an End-to-End ML Model in Snowflake, Getting Started with Snowflake Notebooks on Container Runtime or Introduction to Snowflake Feature Store: docs.snowflake.com How I'd combine it 1. Start with the Hands-On Essentials workshops (Data Warehousing first) to get the basics. 2. Build the Docs agent and use it as your study buddy. 3. Follow a few Quickstarts / Guides for the topic you care about most (Snowpark, Notebooks, ML, Cortex...). They're the fastest way to build something real. 4. Watch a few "Snowflake on Snowflake" sessions for real-world context. 5. Use DataCamp or the Level Up courses to go deeper where you need it (SQL, pipelines, Snowpark, AI). 6. If you want a certification, look at the SnowPro Core prep material. Also: Snowflake Expedition (free virtual event) runs November 3–6, with technical tracks if you want live content. What did I miss? Which resources helped you most? Happy to hear your favorites, especially for Cortex / AI topics. P.S. If you want to work towards badges in a structured way, these two landing pages are a good starting point: - achieve.snowflake.com - developerbadges.snowflake.com