Agentic AI, Security, Governance & Operations

Microsoft via Coursera

Go to class Write review

Overview

Google, IBM & Meta Certificates – 40% Off

One Coursera Plus subscription covers most Professional Certificates on Coursera.

Unlock All Certificates

This course is designed for cloud architects, AI engineers, and technical leads who have completed foundational Azure AI platform training and are ready to take Azure AI solutions from design through production. Learners should be familiar with Microsoft Foundry core platform capabilities before beginning. This course covers the full lifecycle of enterprise AI deployment: agentic system design, security architecture, governance and responsible AI, deployment pipelines, observability, and end-to-end architecture synthesis. You'll evaluate architectural approaches for Foundry-based agents and multi-agent orchestration with MCP tool integration and human-in-the-loop controls, assess security and identity strategies including prompt injection defenses, apply governance and responsible AI frameworks, compare deployment and CI/CD approaches, evaluate observability strategies using Foundry Observability and Azure Monitor, and synthesize these concepts in an end-to-end architecture capstone.

Syllabus

Agentic AI Solutions: Design Agent Tools, Actions, Memory & MCP Integration
- This module develops the skill of designing the core capability architecture of a Foundry-based enterprise AI agent—specifying the tools the agent can invoke, the actions it can execute, how it manages conversation state and memory across turns, and how it integrates with external APIs via MCP—producing a complete agent specification that reflects the functional design of a production enterprise agent.
Agentic AI Solutions: Design Escalation Logic, Guardrails & Human-in-the-Loop Checkpoints
- This module develops the skill of designing the operational control architecture for a production enterprise AI agent—specifying escalation thresholds and human handoff logic, content safety guardrails, and human-in-the-loop checkpoints that help ensure the agent operates safely, within its intended scope, and under appropriate human oversight in high-stakes enterprise environments.
Multi-Agent Architecture: Design Orchestrator-Specialist Patterns & A2A Communication
- This module develops the architectural design skills needed to structure a multi-agent system—defining the orchestrator-specialist topology, assigning agent roles and responsibilities, and specifying the A2A inter-agent communication parameters that govern handoffs. Learners apply these skills to a complex enterprise process that requires coordination among multiple specialist agents.
Multi-Agent Architecture: Design Foundry Workflows with Error Handling & Observability
- This module extends the multi-agent topology design from Module 1 into the operational layer—designing the Foundry Workflow logic that orchestrates agent execution, with error handling, retry logic, and observability instrumentation. Learners apply these skills first through a pre-production architecture review coach dialogue, then through a structured workflow failure diagnosis—identifying production gaps, specifying remediations, and communicating findings to an engineering team as a practicing architect would before any enterprise deployment goes live.
Azure AI Security: Design the Network Security Architecture
- This module develops the skill of designing the network security topology for an enterprise Azure AI deployment—specifying private endpoint configuration, virtual network (VNet) integration, Azure Application Gateway with WAF for public-facing endpoints, and DDoS protection—to help prevent unauthorized network access and protect Azure AI services from network-level threats.
Azure AI Security: Design Identity Controls & Adversarial AI Defenses
- This module develops the skill of designing the identity and access control architecture for an enterprise Azure AI deployment—specifying managed identity for service-to-service authentication, RBAC role assignments for each stakeholder group using least-privilege principles—and designing the adversarial AI defense architecture using Microsoft Defender for Cloud AI threat protection and Azure AI Content Safety to protect against prompt injection and other AI-specific attack patterns.
AI Governance: Unified Access Layer and Policy Enforcement
- This section teaches you to design governance frameworks that provide centralized control over AI model access while enabling appropriate use across the enterprise.
AI Governance: Audit Logging and Compliance
- This section teaches you to design audit logging and traceability controls that meet compliance requirements and enable effective incident investigation.
Responsible AI: Apply Microsoft's Principles to Evaluate AI Solution Designs
- This module develops the evaluative skill of systematically applying Microsoft's six responsible AI principles to a real AI solution design: identifying gaps against each principle and assessing how severe those gaps are in a high-stakes deployment context.
Responsible AI: Design the Governance Framework for a High-Stakes AI Solution
- This module develops the skill of designing a complete responsible AI governance framework—translating identified gaps in an AI solution's responsible AI review into a structured governance architecture with approval workflows, usage policies, compliance controls, audit logging, and human oversight checkpoints.
Deployment Architecture: Evaluate and Select Azure AI Deployment Options
- This module develops the evaluative skill of assessing Azure AI deployment options—managed online endpoints, serverless API deployment, and containerized deployment—against the specific scalability, cost, and latency requirements of a production AI solution scenario, and producing a documented deployment option recommendation that justifies the selection and addresses alternatives.
Deployment Architecture: Design the CI/CD Pipeline and Environment Strategy
- This module develops the skill of designing the full deployment architecture for a production Azure AI solution—specifying the environment strategy, designing the CI/CD pipeline with automated evaluation gates and governance checkpoints, and configuring the managed endpoint deployment—producing the deployment runbook that governs how the solution moves from development through production.
Observability: Architecture and Distributed Tracing
- This module teaches you to design GenAI observability architectures with distributed tracing that provides visibility across the complete request path from user to model and back.
Observability: Dashboards, Alerts, and Continuous Evaluation
- This module teaches you to configure production dashboards, set up continuous evaluation on live traffic, and implement alerting for quality and safety metrics.
Project Module: Agentic AI, Security, Governance, & Operations Architecture
- Learners produce a portfolio-ready Agentic AI, Security, Governance, and Operations Architecture document for a provided enterprise scenario integrating agent design, multi-agent orchestration, security architecture, AI governance framework, responsible AI assessment, deployment strategy, and observability stack into a single coherent production architecture that mirrors the deliverable a practicing Azure AI Solutions Architect would produce for an enterprise agentic AI platform engagement.

Taught by

Microsoft

Reviews

Start your review of Agentic AI, Security, Governance & Operations

Go to class

Taught by

Tags

Never Stop Learning.