Develop AI Apps with Azure AI Foundry Training is a practical, hands-on course designed to help developers and technical professionals build intelligent applications using Azure AI Foundry and modern generative AI technologies. Participants will learn how to design, develop, test, evaluate, and deploy AI-powered applications that use large language models, prompts, enterprise data, and AI services.
Develop AI Apps with Azure AI Foundry Course Overview
This instructor-led course provides practical knowledge and skills for developing generative AI applications with Azure AI Foundry, now documented by Microsoft as Microsoft Foundry. Participants learn how to establish Foundry projects, explore and deploy foundation models, create conversational AI applications, integrate external data and tools, evaluate application quality, and implement responsible AI safeguards.
The course combines essential concepts with guided development activities using the Foundry portal, supported SDKs, model endpoints, and Azure services. By the end of the course, participants will have developed an end-to-end, production-oriented AI application.
Duration 3 Days – 21 hrs.
Learning Objectives
- Explain the capabilities and architecture of Azure AI Foundry.
- Create and configure Foundry resources and projects.
- Explore the model catalog and select models based on solution requirements.
- Deploy and interact with foundation models through secure endpoints.
- Use the model playground to test prompts and model behavior.
- Develop generative AI applications using a supported SDK.
- Build conversational applications using the Responses API or Chat Completions API.
- Apply prompt-engineering techniques to improve response quality.
- Ground model responses using retrieval-augmented generation.
- Extend AI applications with built-in and custom tools.
- Implement secure authentication using Microsoft Entra ID.
- Evaluate AI application quality, safety, and performance.
- Apply content filters, guardrails, and responsible AI practices.
- Add tracing and monitoring to troubleshoot AI applications.
- Prepare an AI application for secure and reliable deployment.
Target Audience
- AI application developers
- Software developers and engineers
- Cloud application developers
- Azure developers and solution architects
- AI engineers
- Data scientists developing generative AI solutions
- Technical consultants
- DevOps and platform engineers supporting AI workloads
- Technical team leaders responsible for AI initiatives
Prerequisites
- Basic knowledge of artificial intelligence and generative AI concepts.
- Familiarity with Microsoft Azure and the Azure portal.
- Working knowledge of Python or C#.
- Basic experience using REST APIs and software development kits.
- Familiarity with JSON, environment variables, and application configuration.
- Basic understanding of cloud identity, authentication, and access control.
- An active Azure subscription with permission to create or use the required Foundry resources.
- A development environment with Visual Studio Code and the appropriate language tools installed.
Course Outline
Day 1 — Azure AI Foundry Fundamentals and Model Development
Module 1: Introduction to Generative AI Application Development
- Generative AI concepts and common business scenarios
- Foundation models, prompts, tokens, and context windows
- Generative AI application architecture
- Key considerations when planning an AI solution
- Introduction to responsible AI
Module 2: Exploring Azure AI Foundry
- Azure AI Foundry capabilities and components
- Foundry resources, projects, hubs, and project endpoints
- Navigating the Foundry portal
- Developer tools, SDKs, and APIs
- Resource providers, subscriptions, and regional availability
- Roles and access permissions
Module 3: Creating and Configuring a Foundry Project
- Creating a Foundry resource and project
- Configuring project settings
- Managing connections and dependent resources
- Understanding endpoints and credentials
- Using Microsoft Entra ID authentication
- Protecting secrets and application configuration
Module 4: Selecting and Deploying Models
- Exploring the model catalog
- Comparing model capabilities and limitations
- Reviewing model benchmarks
- Selecting a model for a business scenario
- Understanding deployment options
- Deploying a model to an endpoint
- Managing model deployments and quotas
Module 5: Experimenting with Models and Prompts
- Using the model playground
- Configuring system and user messages
- Controlling temperature, token limits, and other parameters
- Designing effective prompts
- Applying few-shot prompting
- Testing and comparing model responses
Day 2 — Building and Grounding AI Applications
Module 6: Developing with Foundry SDKs and APIs
- Choosing an appropriate SDK and endpoint
- Configuring a development environment
- Connecting an application to a Foundry project
- Authenticating with Microsoft Entra ID
- Working with deployed models programmatically
- Handling application configuration and credentials
- Managing errors, retries, and API limits
Module 7: Building a Generative AI Chat Application
- Designing a conversational application
- Using the Responses API
- Using the Chat Completions API
- Maintaining conversation history and context
- Managing system instructions
- Structuring model output
- Streaming responses
- Creating an interactive chat experience
Module 8: Prompt Engineering and Output Optimization
- Writing clear and reusable prompt instructions
- Applying prompt templates
- Grounding responses with contextual information
- Requesting structured output
- Reducing hallucinations
- Balancing response quality, latency, and cost
- Testing alternative prompt strategies
Module 9: Retrieval-Augmented Generation
- Understanding retrieval-augmented generation
- Identifying suitable grounding data
- Preparing and indexing organizational content
- Creating embeddings and performing vector searches
- Connecting an AI application to indexed data
- Generating grounded responses with citations
- Improving retrieval relevance
- Applying access controls to enterprise data
Module 10: Extending Applications with Tools
- Understanding tool-enabled AI applications
- Using file search
- Using code interpreter capabilities
- Defining and calling custom functions
- Connecting models to external APIs
- Handling tool-call requests and results
- Controlling tool access and permissions
- Designing reliable tool execution workflows
Day 3 — Quality, Safety, Monitoring, and Deployment
Module 11: Developing AI Agents
- Understanding AI agents and agent use cases
- Comparing chat applications and agents
- Creating an agent in Foundry
- Configuring agent instructions
- Adding knowledge sources and tools
- Managing agent conversations and threads
- Integrating an agent into an application
- Applying agent design best practices
Module 12: Evaluating AI Applications
- Defining application quality criteria
- Creating representative evaluation datasets
- Evaluating relevance, coherence, groundedness, and fluency
- Comparing prompts and model configurations
- Reviewing evaluation results
- Identifying failure patterns
- Using evaluation findings to improve the application
Module 13: Responsible AI, Safety, and Security
- Applying Microsoft responsible AI principles
- Identifying potential harms and misuse scenarios
- Configuring content filters and guardrails
- Defending against prompt injection
- Protecting sensitive and personal information
- Applying least-privilege access
- Securing endpoints, keys, and project resources
- Establishing human oversight and escalation controls
Module 14: Tracing, Monitoring, and Optimization
- Adding application tracing
- Inspecting model and tool interactions
- Troubleshooting application behavior
- Monitoring token consumption, latency, and errors
- Managing cost and model utilization
- Optimizing prompts and application calls
- Planning operational monitoring and alerting
Module 15: End-to-End AI Application Development
- Defining an AI application scenario
- Selecting and deploying an appropriate model
- Designing application prompts
- Implementing conversational interactions
- Adding grounding or tool integration
- Applying authentication and safety controls
- Evaluating and refining the solution
- Preparing the application for deployment
