Snowflake By The Bay

We’re always excited to see our community grow—not just in numbers, but in the diversity of ideas, tools, and brilliant minds. Year after year, our conferences and meetups have introduced us to incredible new collaborators, and we’ve been lucky to build lasting relationships along the way. This year, we’re especially thrilled to welcome Snowflake to our stage.

RSVP

Making their first joint appearance, Danica Fine and Josh Reini, who are teaming up to deliver a talk “Donut Kill My Vibe: Enriching RAG with Apache Iceberg™ for Sweet, Structured Retrieval.” that’s equal parts clever and technically compelling.

Most RAG (Retrieval-Augmented Generation) systems lean on unstructured text, but Danica and Josh make the case that structure can make all the difference—especially when users ask for something specific. Their demo centers around DonutBot, an AI assistant that interprets vague but very human requests like “something sweet and seasonal with sprinkles, under 400 calories,” and turns them into real product matches. The key is combining vector search over product descriptions with structured filters from Apache Iceberg™, which acts as the source of truth for donut metadata. Add in TruLens to evaluate groundedness and relevance, and you’ve got a system that’s accurate, adaptable, and a lot harder to hallucinate with.

While this is Snowflake’s first year officially on our stage, it’s not the first time they’ve been part of the conversation. In 2023, Sherin Thomas delivered one of the most practical talks of the event: Recipe for Building a Discoverable and Governed Data Platform.

Sherin shared how the team at Chime tackled the challenge of discovery and governance at scale—using Datahub to continuously pull metadata from across their stack. Snowflake was one of the most critical platforms mentioned, alongside tools like Looker, Airflow, and Kinesis. While not the focus, its presence in that ecosystem spoke to its role in complex, fast-moving data environments.

What Snowflake Is Focused on Right Now

Snowflake continues to invest heavily in its AI Data Cloud, a unified platform that brings together data engineering, analytics, machine learning, applications, and cross-team collaboration. It’s built to break down silos and help organizations move faster by centralizing everything in one place.

The platform is fully managed—no infrastructure to maintain, no software to install—and runs seamlessly across major public clouds, offering a consistent experience regardless of environment. It supports a wide range of workloads: from data warehousing and data lakes to AI/ML, cybersecurity, and app development.

And the platform is evolving quickly. In the last few months, Snowflake has rolled out several key updates:

• Next-gen data ingestion with Snowpipe Streaming, improving streaming performance with predictable, usage-based pricing.

• Native app release channels, giving developers better control over app deployment workflows.

• Apache Iceberg™ support for structured, high-performance table operations—especially useful in hybrid AI systems like RAG.

• Enhanced task orchestration, including support for streams on directory tables and shared data.

• Continued evolution of Snowpark, Snowflake’s engine for running data engineering and ML workflows directly within the platform, now with smarter scheduling, Python caching, and data skew handling.

  
  

Snowflake’s growing presence in our community reflects the company’s broader shift—from data warehousing to a unified AI Data Cloud powering the next generation of intelligent, scalable systems.

RSVP