Cloud Data Driven

Get ready to explore the next generation of SQL innovation across Azure SQL, SQL Server, and the brand‑new SQL in Microsoft Fabric. We’ll break down what’s changing, why it matters, and how you can start taking advantage of these capabilities immediately.

🔹 Azure SQL & SQL on Fabric See how the newest features—database mirroring, real‑time analytics, and intelligent performance insights—unlock scenarios that were impossible just a year ago.

🔹 SQL Server: What’s Coming Next Preview the upcoming SQL Server release with a focus on performance enhancements, security improvements, and operational efficiency that will reshape how teams build and manage data systems.

🔹 Copilot for Data Professionals Watch how Copilot transforms the way we query, manage, and reason about data, accelerating productivity and simplifying complex tasks across the SQL ecosystem.

If you’re working in data engineering, analytics, or database administration, this session will give you practical, future‑ready insights you can apply right away.

Register now and join us—don’t miss this deep dive into the future of SQL.

As AI becomes embedded in cloud data platforms, organizations are encountering failure modes that differ fundamentally from traditional deterministic systems. In contrast to binary failures, AI-augmented pipelines often degrade silently, with quality erosion emerging gradually and compounding over time.

This talk explores key failure patterns in production AI data systems. Drawing from machine learning systems research, we examine how data and concept drift can persist undetected as models continue producing outputs despite shifting feature distributions. The risk intensifies in chained pipelines, where probabilistic errors compound across stages—for example, when upstream model inaccuracies propagate and amplify downstream.

We also analyze non-deterministic inference behavior, which complicates reproducibility, auditability, and root cause analysis in cloud environments. The session highlights risks of AI-generated data contamination, where synthetic outputs are mistakenly treated as ground truth, accelerating feedback loops and long-term model degradation.

At the infrastructure level, we discuss challenges such as non-linear inference cost scaling, observability gaps that mask semantic failures, and automation complacency that reduces human oversight.

The talk concludes with practical design principles for cloud data systems, including metadata-first architectures, explicit trust boundaries, and human-in-the-loop checkpoints to build resilient, auditable, and trustworthy AI-driven pipelines.

SQL Server 2025 is more than just an upgrade—it’s a leap forward. From becoming an AI-ready database to delivering modern development tools, this release is packed with innovations that redefine performance, scalability, and productivity.

In this session, you’ll see demos and practical examples showcasing a wide range of enhancements, including:

Improved JSON handling
T‑SQL REST API integration
Expanded T‑SQL functionality, including fuzzy string matching and regular expression support
Time‑bound Extended Events
Change event streaming for real‑time data integration
Mirroring in Microsoft Fabric
Updates to change tracking and columnstore indexes

Whether you’re a developer, DBA, or architect, this session will help you understand where SQL Server 2025 fits into your broader data platform strategy. Join us to explore what’s new and improved in SQL Server 2025!