Presentation information coming soon!

At Snowflake, we’ve been pursuing how to use Agents and AI to let any business user talk to their data. Enterprise data isn’t always tidy—and AI agents need more than great retrieval to drive real value. In this session, we’ll share what we’ve learned at Snowflake about enabling agents that deeply understand and reason over structured business data. We’ll cover challenges like navigating messy schemas, generating trustworthy SQL, ensuring consistency in definitions like “revenue,” and making the agent’s process visible to non-technical users. Whether you’re scaling agent use across departments or starting to integrate them into core business workflows, you’ll leave with strategies to make agents effective, reliable, and trusted partners in the enterprise.

GPU cluster reliability is a growing challenge as AI models and the clusters that host them grow to unprecedented scale. Insidious errors such as Silent Data Corruptions (SDCs) are particularly difficult to address due to their highly elusive and non-deterministic nature, and their effect on large-scale LLM training and inference is poorly understood. In this talk, we will present how NVIDIA is leveraging its deep expertise in GPUs and AI to holistically tackle this challenge from silicon to data centers. We will go over the work we are doing to improve our understanding of these complex errors and their effect in real world at-scale AI cluster deployments, and the solutions we are developing to help researchers, cluster builders, and the industry protect against SDCs.

We will explore the world of reinforcement learning, post training and what it’s like to build a startup on open models.

In a future world where agents interact with billions of users, many of these agents will also have to interact with data querying tools to provide answers grounded in facts. As enterprise data analytics is rapidly moving towards open table formats like Apache Iceberg, these agents need to be able to speak to Iceberg-based data. Once agents can speak Iceberg, they gain an advantage - they become portable. They can run in public clouds, locally on a laptop (during development), or on-premise, accessing enterprise data that can't be moved to public clouds. Portability is important because, while Nvidia GPUs dominate in the cloud, the GPU stack looks different on-premises and on consumer hardware. In this talk, we will discuss how Apache Iceberg tooling and portable application runtimes make agents grounded in facts and enable them to run across different GPU stacks and deployment models.

Meta manages a large-scale data warehouse where security is a critical component. Every day, teams across Meta are tasked with managing access to the data they oversee and obtaining access to data through internal data products. In this talk, we delve into the challenges of managing internal data access at Meta's scale and its growing complexity. We will also share how we developed an agentic solution to empower both data users and data owners in addressing these challenges.

As LLM applications evolve into multi-agent systems and power complex decision-making workflows, the ability to observe and debug their behavior becomes a core engineering challenge. These systems are dynamic, non-deterministic, and increasingly reliant on external tools and APIs making traditional monitoring approaches insufficient. At Fiddler, we've worked with enterprise and federal teams deploying LLMs at scale, and what we’ve consistently seen is the absence of effective observability creates blind spots that delay iteration and introduce risk. In this talk, we will introduce Agentic Observability, a set of techniques and infrastructure to monitor production LLM systems. We will walk through how we trace agent reasoning and tool usage in structured form, apply Fast Trust Models to evaluate output quality beyond token-level accuracy, and monitor shifts in behavior using statistical and embedding-based methods. We will also share how we enable integration testing for agent workflows by simulating decision paths and validating semantic intent, all while operating under the scale and latency constraints of modern AI stacks. This work bridges AI science, platform engineering, and real-world GenAI deployment. We will highlight engineering lessons learned from high-scale environments, and how these observability tools are helping teams move faster, catch failures earlier, and build AI systems that can be trusted in production.

Join us as we delve into the high-level design and architecture that enabled the creation of Ray-Ban Meta, a best-in-class AI wearable used by millions. To make AI truly useful, it must be seamlessly integrated into our daily lives, providing reliable and high-performance capabilities. However, achieving this requires overcoming real-world physical limitations through innovative engineering and model design.

In this talk, we'll explore how a user-centric approach drove the development of a complex architecture that harmoniously brings together multiple components to meet the unique needs of our users. From running models directly on the frames to optimizations informed by user behavior, we'll break down the key elements that have made Ray-Ban Meta a game-changer in the world of AI wearables.

If the future is agentic, what does this mean for Infrastructure?