Snowflake Developers

What if your data scientists could determine your most impactful ML features with just natural language? In this demo, we use Cortex Code inside Snowflake Notebooks to: - Identify which features actually drive churn prediction - Flag correlated signals that can distort results - Surface fragile features that may hurt model reliability - Suggest next steps like feature pruning or model changes No manual analysis. No guesswork. Just fast, explainable insights directly where your data lives. See how agentic ML in Cortex Code is transforming ML workflows and accelerating production: https://lnkd.in/g_hsYA7J

This is your inside look at how Snowflake's own Product Data Science team uses Cortex Code in their daily work. Hosted by James Cha-Earley Tyler Richards and Zachary Blackwood will feature real workflows with real outputs! (not demos) Workloads like: pipeline engineering, ad-hoc investigations, recurring analyses: tasks that used to cost hours now happen in a handful of prompts. Join the livestream on Tuesday, March 31st to learn: 👉 How Snowflake's data science team actually uses Cortex Code day-to-day 👉 How multi-step workflows collapse into single prompts, with validation and documentation built in 👉 Real lessons from months of internal use, plus live Q&A with the team!

What does it actually look like when an engineer takes a Jira ticket through dbt, Snowflake, semantic views, and a React frontend in a single terminal session? Join me with Trent Foley and Jonathan Gard from evolv Consulting as they demo the workflow live and we dig into the real questions: where AI fits in analytics engineering today, what has to be true about your data architecture before any of it works, and why engineering discipline matters more with AI than without it.

Exciting news from Snowflake AI Research! The new DeepSpeed Ulysses implementation for Hugging Face (HF) developed by our team is now fully integrated with Hugging Face Accelerate, the Transformers Trainer, and the TRL SFT Trainer. DeepSpeed Ulysses enables efficient training with ultra-long sequences by leveraging aggregate GPU memory across multiple devices — all while keeping communication overhead minimal. If you're working with long-context models, this is a big unlock. Learn more about how to leverage DeepSpeed Ulysses: https://lnkd.in/gNHDFTDX

Thinking about migrating to Snowflake Postgres but want to minimize downtime? Postgres' built in logical replication makes platform migrations much less painful. See our quickstart guide on using logical replication to migrate from one platform host to another: https://lnkd.in/gW8XgDCC

Apache Iceberg™ has conquered batch analytics. What’s next? At Iceberg Summit, Russell Spitzer breaks down what V3 unlocked. Think vendor-neutral support for semi-structured and geospatial data, plus more efficient deletes. But he also gets into what’s still missing for AI and streaming workloads. This keynote looks ahead to V4 proposals like One File Commits, Improved Column Statistics, and Columnar Metrics, and what they unlock for builders. Watch the keynote in person or virtually on April 8 at 9:00 AM PT to see where Iceberg goes next: https://lnkd.in/gdf32zj2

Accelerate the entire machine learning lifecycle with agentic ML from Cortex Code You can plan, build, and run production-ready ML pipelines using natural language directly in Snowflake. Focus on the business problems, not the pipelines and tooling. With Cortex Code, you can: -Build and deploy high quality ML pipelines with built-in MLOps - Easily improve model performance and prediction accuracy quickly using natural language - Automate the tedious work so your data scientists can focus on higher impact initiatives See what's new: https://lnkd.in/gfQgzsQQ

The latest and greatest for developers in Cortex Code: 1️⃣ Cortex Code is now GA in Snowsight, bringing a persistent AI assistant directly into your workflows 2️⃣ Native Windows support for Cortex Code CLI expands access across dev environments 3️⃣ Agent Teams let you break down complex, multi-step work into parallel, coordinated tasks 4️⃣ New agent skills for Streamlit, Openflow, cost intelligence, ML, and more standardize how you build on data See what's new: https://lnkd.in/gvvVtJ6x 

The data landscape is undergoing a fundamental shift as organizations adopt open lakehouse architectures built on Apache Iceberg™. At the core of that shift is interoperability. You can use any engine to write to your Iceberg tables and query them from any engine. But even simple things like identifier casing can break that promise. A table created in one engine may not resolve in another. In this post, we break down how Snowflake solves this with CATALOG_CASE_SENSITIVITY in catalog-linked databases: - CASE_INSENSITIVE for Hadoop-style catalogs (Glue, Unity) - CASE_SENSITIVE for ANSI SQL environments (Polaris) Learn more: https://lnkd.in/gpKx2i4U

Get started with Snowflake DCM Projects. Snowflake DCM (Database Change Management) Projects let you define your Snowflake infrastructure as code using SQL-based definition files. In this quickstart, you'll learn: - How DCM Projects define Snowflake infrastructure as code - How to structure a DCM Project with a manifest and definition files - How to use Jinja templating to parameterize definitions across environments - How to plan (dry-run) and deploy changes using the Snowsight Workspaces UI - How to attach data quality expectations to objects using Data Metric Functions Get started: https://lnkd.in/g-4YQyi9