🌊 Data in Motion: Streaming & CDC for AI & Agents
Agents only adapt as fast as the data reaching them, making the streaming stack underneath critical. Join VeloDB, Redpanda, and Datastrato to learn how teams are building that stack: ingesting at scale, moving data with low latency, and querying it in real-time.
🌊 Data in Motion: Streaming & CDC for AI & Agents
When: 🕠 5:30–8:00 PM (🍽️ dinner included)
When: 🕠 5:30–8:00 PM (🍽️ dinner included)
⚡ Your agent is only as smart as its context.
To achieve an enterprise-ready real-time data stack for agents, an organization needs to look at data movement end-to-end: a stream that moves data from operational sources to analytical targets, a database that can ingest and serve that stream at the speed agents demand, and a way to govern the data as it lands.
Join VeloDB, Redpanda, LaserData and Datastrato for a working session on building that streaming stack: ingesting at scale, moving data with low latency, querying it the second it arrives, and governing it the moment it lands, so your agents always run on the freshest context.
Three short talks, 🍲 dinner, and 🍻 conversation in SF!
✨ Speakers
Real-Time Data Architecture for the Agentic Era
🎤 Peter Corless — Principal Product Marketing Manager, Redpanda
Discover how enterprises are building enterprise-scale agentic AI applications. These real-time responsive systems require several components in their data architecture: event-driven data streaming, real-time analytics engines, and AI-centric application frameworks for governance, trust, and explainability.
Closing the Governance Gap Between the Stream and the Lakehouse
🎤Mark Hoerth — Product Lead & Solutions Architect, Datastrato
Streaming and analytics keep converging, but governance usually doesn't follow the data across the seam. A topic lives in one world with its own access model; the table it becomes lives in another. This talk shows how an open catalog can span both. Using Redpanda's broker-native Iceberg Topics to turn a live stream into an Apache Iceberg table with no ETL, and Apache Gravitino as the catalog of catalogs holding the topic and resulting table in a single metalake, we'll govern data the same way before and after it lands. The session ends with a live walk from produced records to a governed, queryable Iceberg table under one consistent policy. The takeaway: your governance boundary doesn't have to break where your streaming engine hands off to your lakehouse.
Fast In, Fast Out: A Real-Time Analytics Stack with Apache Iggy (Incubating) and Apache Doris
🎤 Kranti, Founder and CEO, LaserData & Kevin Shen, PPM, VeloDB
This talk pairs two open-source projects to build a real-time analytics stack covering both fast ingest and fast queries: Apache Iggy (incubating), LaserData's Rust message-streaming platform that moves high event volumes at very low latency, and Apache Doris (the database behind VeloDB), which serves sub-second queries on fresh data under heavy concurrency, updates, and joins. After a brief intro to each project, the session walks through a Rust-native Iggy sink connector that streams events directly into Doris with no JVM or intermediary systems, then closes with a live demo pushing a real workload through Iggy into Doris and running analytics as the data lands—leaving attendees with an understanding of why real-time analytics needs both halves, how the connector wires them together, and when and how to build the stack themselves.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article