​Apache Iceberg™ Meetup Bay Area! 🧊❄️

​​Join the Iceberg Community on Wednesday, January 21st at the Silicon Valley AI Hub in Menlo Park for an Iceberg meetup!

​​​Connect with fellow enthusiasts, share insights, and dive into the latest developments in the Apache Iceberg™ ecosystem! Whether you're a seasoned pro or new to Apache Iceberg, this meetup is the perfect place to exchange ideas and spark innovation.

​​​Agenda

​​​5:30 PM - 6:15 PM: Networking and Dinner

​​​6:15 PM - 8:15 PM: Welcome Remarks & Presentations!

​​​8:15 PM - 9:00 PM: More Networking

​​The event will focus on use cases around and innovations in Apache Iceberg (https://iceberg.apache.org/)

​​Agenda

​Talk 1: Apache Iceberg as a Versioned Data Backbone for ML & AI, An Phan @ Hippo Harvest

​As ML and AI systems move into production, data reproducibility becomes as critical as model reproducibility. Many teams still rely on ad hoc snapshots, fragile pipelines, or manual conventions that quietly break trust over time, especially in systems connected to sensors or physical environments where data cannot be easily recreated.

This talk will show how Apache Iceberg can be used as a versioned data backbone for ML and AI workflows by treating snapshots and time travel as first class primitives rather than debugging tools.

​Talk 2: Are We There Yet, or Just at the Tip of the Iceberg? Amy Chen @ ClickHouse

​After 2 years of hype—and a continued rush to support and adopt it for analytics, streaming, and AI—the question I want to ask is: have we hit the sweet spot of Iceberg yet?This talk takes a pulse check on how Iceberg actually looks, past the vendor articles, to the lens of customers. With daunting complexity, operational overhead, and performance tradeoffs, is Iceberg ready for the masses? We’ll look at where Iceberg adoption commonly stalls, why “checking the box” isn’t the same as real success, and what’s missing between early wins and sustained impact. Based on real customer patterns, we’ll close with how ClickHouse is helping teams bridge this gap—making real time Lakehouses possible and more reliable as demands continue to rise.

​​Talk 3: Observability for AI Agents, Weimo Liu @ PuppyGraph

​Observability has long been a foundational challenge in modern distributed systems. By collecting telemetry data such as logs, metrics, and traces, engineers gain visibility into system behavior and performance. In traditional microservices environments, execution is largely request-driven, service topology is mostly static, and sampling is commonly used to manage data volume. Observability data is typically analyzed using SQL-based analytics engines over structured or semi-structured datasets.

​With the rise of AI agents, observability faces a fundamental shift in what traces represent. Traces are no longer confined to request paths across predefined services. Instead, agent traces capture the runtime execution of a request as it unfolds: a sequence of model invocations, tool calls, and orchestration steps selected dynamically based on context. At execution time, it is often impossible to know in advance whether an agent will invoke another agent, call a specific tool, or terminate after a single step. Unlike traditional systems, AI agents do not operate over a fixed call graph. Each request may produce a distinct execution structure, causing trace data to form dynamic, data-dependent execution graphs rather than stable paths through a known topology.

​These changes introduce two major challenges for observability systems: a dramatic increase in data scale and continuously evolving graph structures that cannot be fully modeled in advance. In this talk, we demonstrate how Apache Iceberg, combined with a graph engine, provides a clean, flexible, and cost-efficient foundation for addressing these challenges, enabling scalable observability for AI agent-driven systems.

​​Talk 4: Revisiting Apache Iceberg Metadata Structure and Storage, Steven Wu @ Snowflake

​Since the inception of Apache Iceberg, its metadata structure and storage has remained unchanged. Over the years, we have learned what worked well and what didn’t. A few V4 proposals are revisiting some of those earlier design choices and improving the metadata structure and storage. In this talk, we are going to learn what problems we are trying to address and how we are solving those problems with the new V4 proposals.

​About PuppyGraph

​​PuppyGraph is the first and only real time, zero-ETL graph query engine in the market, empowering companies to transform existing relational data stores into a unified graph model in under 10 minutes, bypassing traditional graph databases' cost, latency, and maintenance hurdles.

​​​💬 Join PuppyGraph Community Slack

​​​📚 Check out PuppyGraph Engineering Blog

​​​📲 Follow PuppyGraph on LinkedIn & Twitter

​​​🖥️ Subscribe to PuppyGraph YouTube

​​💾 Download PuppyGraph Forever Free Developer Edition (no form & no payment required)

​About Snowflake

​​​Snowflake makes enterprise AI easy, efficient and trusted. More than 10,000 companies around the globe, including hundreds of the world’s largest, use Snowflake’s AI Data Cloud to share data, build applications, and power their business with AI. Snowflake provides native support for Apache Iceberg™ and Apache Polaris™ (incubating).

​​​📚 Check out how Snowflake can power your open data lakehouse

​​​📲 Follow Snowflake on LinkedIn & X

​​​🖥 Subscribe to Snowflake Developers YouTube

​​​❄️ Start your 30-day free Snowflake trial which includes $400 worth of free usage

​About ClickHouse

​​Established in 2009, ClickHouse leads the industry with its open-source column-oriented database system, driven by the vision of becoming the fastest OLAP database globally. The company empowers users to generate real-time analytical reports through SQL queries, emphasizing speed in managing escalating data volumes.

​​​📚 Get started on ClickHouse Github

​​💬 Join the ClickHouse Slack Channel

​💜 We’re hiring, join our team!