Dennes Torres

A full-day advanced training on building production-grade AI agents with Azure AI Foundry. The session starts with Model Context Protocol (MCP), showing how to build an MCP server using Azure Functions and test it with the MCP Inspector. Participants then create a Foundry agent linked to the MCP server, reuse the same MCP backend in Copilot Studio with a small demonstration, and implement knowledge bases backed by Azure AI Search using hybrid retrieval combining vector, keyword, and semantic search with model-controlled query refinement. The day concludes with an intent-driven workflow that routes requests between RAG-based knowledge retrieval and structured database access in a single agent.

This session explores how they simplify medallion architecture creation into a low-code — and in some cases near no-code — experience, significantly changing how data engineering pipelines are built. You’ll learn how the feature works internally, its limitations, where it fits best, and how to build Materialized Lake Views to streamline and modernize your data pipelines.

Get started with the new Foundry ecosystem from Microsoft and learn how to build your first AI-powered workflow. This session introduces the core concepts of Foundry, explains how the platform works, and walks through creating an MCP server as an Azure Function and connecting it to an agent. You’ll also learn how Foundry evaluates agent performance and how to apply guardrails to ensure your solutions remain safe, reliable, and aligned with organizational standards. A fast-paced, practical introduction for anyone ready to begin building with Foundr

This session explores modern RAG architectures in Foundry, explaining when to use vector-based AI Search indexes versus the new Knowledge Base concept, how they work internally, and how agents can leverage each approach for better grounding and retrieval. Attendees will leave with practical guidance on choosing the right RAG pattern for agent-driven solutions.

SQL Server 2025 introduces Change Event Stream, enabling real-time capture of inserts, updates, and deletes as an event stream. Combined with Microsoft Fabric, this allows transactional data to flow continuously into Eventstream, Lakehouse, and Real-Time Analytics without polling or batch ETL. In this session, you’ll learn how Change Event Stream works, how it differs from CDC and Change Tracking, and how to build low-latency, event-driven architectures that power real-time dashboards, alerts, and modern data applications bridging transactions and analytics.

This session demonstrates a practical marketing solution for review sentiment analysis, showing how Fabric Shortcut Transformations and AI Shortcut Transformations simplify data pipelines, apply AI directly to data, and deliver insights faster. Ideal for data professionals and marketing innovators looking to unlock business value with scalable, reusable techniques.

Discover how SQL Server 2025 transforms data querying by combining native vector support with deep AI model integration. This session shows how semantic search goes beyond keywords, enabling meaning-based queries and more relevant results through vector similarity. Learn how to invoke large language models directly from SQL Server to enrich queries with AI-driven context, summaries, and insights—securely and efficiently. With simplified development compared to earlier Azure SQL approaches, SQL Server 2025 empowers developers and data professionals to seamlessly blend vectors and AI in a single, powerful query experience.

A full-day advanced training on Microsoft Fabric Lakehouse maintenance and performance. The session provides a deep explanation of Delta internals, why OPTIMIZE and VACUUM are required, and how to implement maintenance at scale across multiple lakehouses. Attendees learn how to diagnose performance using lakehouse system tables, map heavy queries back to semantic models and reports, and plan critical configuration choices such as V-Order. The training also covers partitioning, Z-Order, and PySpark query practices for consistent, predictable performance.