Adrien Treuille | Director, Product Management
AI is moving fast... Here are the 10 sessions that I’m most excited about at Summit this year. I can’t wait to hear some of the world’s top AI researchers peer into the future, watch live demos of some of the most magical new technologies in Snowflake, and see how Snowflake is powering AI-based cognition across some of our biggest customers.
Sessions
What’s New: AI Research and Development at Snowflake, WN302
Explore the exciting world of AI research and innovation happening right here at Snowflake! Join us to get firsthand insights into the latest developments in large language models (LLM) straight from the experts.
Understanding Users Through Video Transcripts with Snowflake Cortex AI, AI108
For Vimeo, user-uploaded videos contain rich data insights into user behavior, identity and their reasons for using the Vimeo platform. In this talk, Vimeo will demonstrate how the company uses Snowflake Cortex AI to extract valuable insights from video transcripts. This process enables Vimeo to identify the most profitable video categories to inform its content and marketing strategies. Join this session to also learn how Vimeo created an interface for account representatives to gain on-the-fly context about their customers to enhance customer service. (Please note: Theater sessions are first-come, first-seated.)
Talk to Your Data: The New Era of Data Analysis Powered by AI, WN102
Join us for an exciting journey into the future of AI and data. Discover how Snowflake Copilot, a breakthrough AI-powered SQL assistant, turbocharges data analysis while maintaining robust security and governance. In this session, we’ll explore Copilot's transformative impact on enterprise analytics and offer a sneak peek into what’s next: enabling business users to converse with their data directly. Prepare to unlock new levels of efficiency and insight!
How Zoom Talks to Its Data with Snowflake Cortex AI and RAG, AI308
This session will focus on interacting with data on Snowflake using Cortex AI, large language models (LLMs) and retrieval augmented generation (RAG). We will demonstrate how leveraging advanced natural language processing capabilities can empower business leaders to engage in intuitive conversations with their data.
An Overview of Generative AI at Snowflake, AI109
This session provides a roadmap to Snowflake’s current generative AI (gen AI) capabilities, and shows how they can empower your applications. We’ll provide an overview of how Snowflake Cortex AI — including Snowflake Arctic and Arctic Embed — empower customers to build cutting-edge large language model applications with ease. Join to hear our perspective, current capabilities, roadmap and vision for enterprise AI.
What's New: Building AI Chat Experiences with RAG in Snowflake Cortex, WN202
Looking to bring more value out of documents and other text data? Join us to learn about the latest serverless LLM functions and how you can get up and running with RAG-based AI chat experiences within minutes inside Snowflake.
AI Research Visionaries Part 2: Drs. Andrew Ng And Hanna Hajishirzi, AI215
Amidst the dizzying pace of AI advances, it is easy to lose focus on the big questions: What is the future of human learning in the era of machine learning? How will our children be taught, and what will they learn? What will companies and other forms of human organization look like in 10 years? How might AI transform the future of work? Snowflake is proud to present a panel of three of the world's most prominent AI researchers: Drs. Andrew Ng and Hannah Hajishirzi. The moderator, Dr. Adrien Treuille — former professor at Carnegie Mellon, former CEO of Streamlit and now Director of AI Product at Snowflake — will guide our exploration into these profoundly important questions, drawing on the vast experiences and insights of our distinguished guests.
How to Build and Scale an Entirely Open Source AI Stack, AI119
If you're looking to interact with LLMs today, you have two paths forward: fine-tune an open source LLM or leverage an existing model API. In this session, you'll learn when it's appropriate and useful to leverage open source LLMs, as well as how to build and scale a fully open source and reproducible gen AI application. This session will cover the role open source plays in model bias awareness, code security and observability, long-term infrastructure considerations and attracting top talent. You'll also learn how to assemble your own open source gen AI stack and what to consider in your architecture. We'll pull back the curtain on some critical, but less talked about processes in LLM development, such as how our team accelerates pre-training. We'll also show you how to reproduce this yourself using open source tools such as Modin and Streamlit. Please note: Theaters are first come, first seated. If this session is full, please sign up for the overflow session here.
The Future Of Trust In LLMs, AI121
In this session, Richard Socher, CEO and founder of You.com and AIX Ventures will share insights from his journey over a decade of research in AI and natural language processing — from the invention of prompt engineering while serving as Chief Scientist at Salesforce to founding You.com, the first AI assistant to integrate a large language model (LLM) with live web access for accurate, up-to-date answers with citations. Richard will discuss how to tackle the biggest challenges facing LLMs, from hallucinations to generic responses. Join to gain insight into the potential for these advancements to be adopted by other LLM-based platforms. A Q&A session will follow. Please note: Theaters are first come, first seated. If this session is full, please sign up for the overflow session here.
Snowpark Container Services and Bio LLMs for Accelerating Drug Discovery, AI106
Generative AI is accelerating our ability to understand all aspects of drug discovery, including predicting the 3-D structures of proteins. In this demo, we’ll explore how to run the protein folding model from BioNeMo by NVIDIA using enterprise-grade data in Snowflake with Snowpark Container Services. We will also explore how to leverage custom protein embedding models to perform similarity searches to find similar proteins, and compare their structures to see if the sequence similarity resulted in structural similarity. For this we will leverage Uniprot RAW embeddings made available by EMBL and the protT5 model for obtaining the protein embeddings. Please note: Theaters are first come, first seated.