Uploaded July 2026 | Updated September 2026, 2 weeks ago
En esta sesión, desglosaremos por qué los fundamentos y herramientas sólidas de DevOps son aún más importantes a medida que los agentes de programación y operaciones impulsados por IA se convierten en parte de los flujos de trabajo cotidianos. Verás cómo los agentes de desarrollo de IA modernos, como GitHub Copilot Cloud Agent y los agentes de operaciones de IA como el agente de Azure SRE pueden acelerar drásticamente la entrega cuando tus pipelines, pruebas y entornos están diseñados para soportarlos. Nos centraremos en lo que los desarrolladores deben tener implementado para integrar de manera segura y efectiva estos agentes en flujos de trabajo de CI/CD del mundo real, sin sacrificar la calidad, confiabilidad o control.
#microsoftreactor #learnconnectbuild
[eventID:27332]
En esta sesión, desglosaremos por qué los fundamentos y herramientas sólidas de DevOps son aún más importantes a medida que los agentes de programación y operaciones impulsados por IA se convierten en parte de los flujos de trabajo cotidianos. Verás cómo los agentes de desarrollo de IA modernos, como GitHub Copilot Cloud Agent y los agentes de operaciones de IA como el agente de Azure SRE pueden acelerar drásticamente la entrega cuando tus pipelines, pruebas y entornos están diseñados para soportarlos. Nos centraremos en lo que los desarrolladores deben tener implementado para integrar de manera segura y efectiva estos agentes en flujos de trabajo de CI/CD del mundo real, sin sacrificar la calidad, confiabilidad o control.
#microsoftreactor #learnconnectbuild
[eventID:27332]
![Production-Ready AI Systems: Security, Evaluation & Data Platforms
Modern AI systems require more than powerful models—they require security, evaluation, governance, and continuous improvement. This session combines lessons from production AI agent security with real-world LLM evaluation and fine-tuning workflows.
Topics may include prompt injection, tool abuse, memory poisoning, defense-in-depth architectures, custom evaluation frameworks, Azure OpenAI fine- tuning, and practical engineering lessons learned from deploying AI-powered systems.
Key Takeaways:
- Understand security challenges in AI agents
- Learn practical defense patterns for production AI
- Explore LLM evaluation methodologies
- Understand fine-tuning workflows using Azure OpenAI
- Apply production engineering best practices to AI systems
📌 This is part of a series, learn more here: https://aka.ms/ProdReadySystems/series
00:00 Intro & Housekeeping
01:50 Securing the AI Stack: Why AI Security Matters
05:00 The Four-Layer AI Security Framework
06:48 Layer 1: Model Security (Prompt Injection, Jailbreaks & Output Hijacking)
10:55 Layer 2: Agent & Application Security
14:09 Layer 3: MCP & Tool Security Risks
16:06 Layer 4: Infrastructure Security Essentials
17:57 Defense in Depth & Security Checklist
19:42 Resources & Final Takeaways on AI Security
20:43 LLM-Driven Merge Conflict Resolution Introduction
22:21 Why Merge Conflicts Are Still a Major Developer Challenge
28:23 Fine-Tuning LLMs for Merge Conflict Resolution
29:40 Key Fine-Tuning Insights & Lessons Learned
33:31 Evaluating Merge Resolution Models
34:42 Python Tips, Structured Outputs & Development Best Practices
39:00 Handling Large Files & Token Limits
40:46 Q&A Transition
42:07 Agent Security in Practice: Real-World Risks & Attacks
43:26 Hugging Face Security Incident Case Study
47:44 OWASP Top Risks for LLM Applications
54:05 Live Demo: Testing & Defending Against Prompt Injection
56:17 Excessive Permissions, Data Exposure & Supply Chain Risks
59:30 Token Abuse, Rate Limiting & Secure Agent Design
01:01:09 Evaluations, Red Teaming & Reliability Testing
01:03:09 Slides, Q&A & Resources
01:03:46 Closing Remarks & Event Survey
[eventID:27335] Production-Ready AI Systems: Security, Evaluation & Data Platforms](https://i.ytimg.com/vi/Bo0XKu7bZ54/mqdefault.jpg)
![Model Mondays - Spotlight On: MAI Models
At Microsoft Build, we announced a family of seven new models developed in-house at Microsoft AI. These models form a new multimodal ecosystem designed to work across a variety of real-world tasks - with models to support image, voice, transcription, coding, and reasoning.
Join us as we talk to the Microsoft AI models team about whats new in the MAI family, how this works under the hood, and what we can do to use them more effectively for building AI agents.
🔎 Explore the resources: https://aka.ms/model-mondays
📢 Continue the conversation on the Discord: https://aka.ms/model-mondays/discord
📖 Read the Newsletter: https://aka.ms/model-mondays/newsletter
📌 This session is a part of a series! Learn more here: https://aka.ms/model-mondays/rsvp
00:00 Welcome & Stream Introduction
01:05 Model Mondays Returns: Microsoft AI Models Spotlight
03:10 Microsoft AI Vision: Token Efficiency & Trustworthy Models
07:11 How MAI Models Power Microsoft Products and Foundry
11:01 Hill Climbing, Reinforcement Learning & Excel Agent Training
17:20 Building MAI Code Models for GitHub Copilot
22:17 Live Demo: Generating a Web App from Scratch
24:08 Enhancing Existing Apps with AI Coding Assistance
30:10 Sneak Peek: Native Image Understanding in MAI Code 1.1 Flash
34:14 Exploring MAI Image Models, Character Consistency & Latest AI News
54:18 Wrap-Up and Next Weeks Cohere Models Preview
#MSFTReactor #learnconnectbuild
[eventID:27393] Model Mondays - Spotlight On: MAI Models](https://i.ytimg.com/vi/C7SUG4s7aLo/mqdefault.jpg)

![Certifícate SQL+AI (DP-800): Diseña y Desarrolla Soluciones SQL como un Profesional
Esta sesión se centra en la parte de “construir” del DP-800: cómo diseñar y desarrollar soluciones de bases de datos que funcionen en cargas reales y que estén alineadas con lo que evalúa el examen. Recorreremos las habilidades clave del desarrollador, desde elegir los objetos y patrones correctos hasta escribir SQL sólido y mantenible.
💡 Esta sesión es parte de una serie—aprende más: https://aka.ms/datadayslive/sp-y
0:00 Welcome & Introduction
1:50 Meet the Speakers
3:45 Data Days, Certifications & Community Resources
8:49 DP-800 Exam Overview
14:45 Database Design & SQL Platforms
25:30 SQL Programmability Objects (Views, Stored Procedures, Functions & Triggers)
39:47 Advanced SQL Development (CTEs, Window Functions, JSON & Graphs)
46:41 AI-Assisted SQL Development (GitHub Copilot, Fabric Copilot & MCP)
58:43 DP-800 Exam Preparation Tips & Key Takeaways
1:02:04 Feedback Survey & Closing Remarks
[eventID:27403] Certifícate SQL+AI (DP-800): Diseña y Desarrolla Soluciones SQL como un Profesional](https://i.ytimg.com/vi/CPJB_-ZbRZo/mqdefault.jpg)



![Secure by Design: Building Security into Every Commit with GitHub Copilot
Security isnt just the security teams job.
In this episode of New Breakpoint, Michelle Sandford is joined by George Coldham, Cloud Solution Architect at Microsoft, to explore how developers can use GitHub and GitHub Copilot to make security a natural part of their everyday workflow.
Through a live demo, George shows how to identify exposed secrets, detect vulnerable dependencies, improve code quality, and leverage AI-powered tools to address issues before they reach production.
Whether youre an experienced developer or just getting started, youll learn practical ways to build more secure applications while maintaining speed and productivity, and why human accountability remains critical in AI-assisted development.
Learn more about the series: https://aka.ms/NewBreakpoint/r
Check out the resources: https://aka.ms/new-breakpoint/log
00:00 Introduction
03:49 Demo Application Overview
09:21 Security Is Everyones Responsibility
12:23 Discovering an XSS Vulnerability
20:13 Using GitHub Copilot to Fix Security Issues
22:24 Testing and Validating the Fix
26:08 Pull Requests, Reviews & Branch Protection
31:24 Security & Quality Dashboard
33:24 Code Scanning, Secret Scanning & Dependabot
34:33 Copilot Autofix and Security Remediation
37:10 Managing Package Vulnerabilities
44:04 Secret Management & Push Protection
49:38 CI/CD Security Validation
51:20 GitHub Copilot Optimization Tips
55:22 Security-First Development Practices
57:28 Closing & Resources
[eventID:27494] Secure by Design: Building Security into Every Commit with GitHub Copilot](https://i.ytimg.com/vi/DyGumvvyYy8/mqdefault.jpg)
![Build AI-powered soccer insights apps with cloud-native Azure Databases
Learn how to build and deploy AI-powered applications that analyze football match and player data using cloud-native Azure Databases.
Using Premier League-inspired scenarios such as identifying team play styles, analyzing counterattacks, or uncovering player performance patterns, well walk through a reference architecture that pairs operational data in Azure SQL, PostgreSQL, and Cosmos DB with vector search and retrieval to ground your models in real match data.
Attendees will leave with practical guidance and ideas they can extend into their hackathon projects.
📌 This event is part of a series. Learn more here: https://aka.ms/Pre
ierLeagueHack
🔗 Learn more by exploring the resources:
https://aka.ms/AzureCosmosDBnoSQL
https://aka.ms/datadrivenappsqldb
https://aka.ms/azure-database-postgresql
#MicrosoftReactor #LearnConnectBuild #PremierLeagueHackathon
[eventID:27555] Build AI-powered soccer insights apps with cloud-native Azure Databases](https://i.ytimg.com/vi/EQ187ohgCbo/mqdefault.jpg)

[eventID:27507] Beyond Chatbots: Build Your First Production Agent with Microsoft Foundry](https://i.ytimg.com/vi/ExJXHCkeO4A/mqdefault.jpg)
![Designing Better Data Models with an AI Coding Agent
Part of the Modern AI Apps with Azure Cosmos DB Series.
Data modeling is one of the hardest parts of building distributed applications, and its easy to get wrong early. This session explores how an agent can help design containers, partition keys, relationships, and access patterns for Azure Cosmos DB while teaching developers how to validate those recommendations.
Presenters:
Sajeetharan Sinnathurai - Azure Cosmos DB Principal Program Manager - https://www.linkedin.com/in/sajeetharan/
Jay Gordon - Azure Cosmos DB Senior Program Manager https://www.linkedin.com/in/jaygordon0042/
• Start with the Agent Kit — Give your coding agent the context it needs to build Cosmos DB apps the right way - https://aka.ms/CosmosDB-agentkit
• Subscribe to Azure Cosmos DB on YouTube - https://youtube.com/AzureCosmosDB
• Follow Azure Cosmos DB on X - https://twitter.com/AzureCosmosDB
• Follow Azure Cosmos DB on LinkedIn - https://www.linkedin.com/company/azure-cosmos-db/
#azurecosmosdb #aicoding
[eventID:27240] Designing Better Data Models with an AI Coding Agent](https://i.ytimg.com/vi/F2RKDp65WUw/mqdefault.jpg)