Topics of Interest

All submissions will be reviewed by experts in the field. The authors of the submissions are invited to present their session live (online) under the following talk categories: 

Track 1: Real-Time Data for AI  

This track explores how modern AI systems are delivering sub-second intelligence in production. We are looking for talks that dive into the architecture and engineering challenges of building real-time pipelines, deploying online feature stores, and handling high-throughput, low-latency workloads, and building real-time AI systems using LLMs. Share your experiences with event-driven data systems, maintaining data freshness, managing consistency at scale, and the operational lessons learned from serving real-time models in fast-moving environments.

Track 2: External Data for LLMs

This track focuses on the role of external data for LLM-powered applications. How do you use Retrieval-Augmented Generation (RAG) to add examples to your prompts? How do you integrate external data sources in your agents using Model Context Protocol (MCP)? How do you support time-travel for RAG or MCP data sources? Can you create EVALs from historical data or do you only create EVALs using log-and-wait in production services? How do you manage LLM traces and logs? How do you manage the creation of instruction datasets for fine-tuning LLMs?

Track 3: Lakehouse Data for AI

The rise of Open Table Formats (Lakehouse Tables) has changed how we store and retrieve tabular data for AI systems. This track is dedicated to exploring the convergence of Lakehouse data with AI systems. We invite talks on how teams are integrating Lakehouse data with training and inference infrastructure. If you are rethinking how Lakehouse data fits into your AI stack, we want to hear from you.

Important Dates

Submissions Deadline: July 6th, 2025

Results: July 14th, 2025

Conference Date: October 14th, 2025