× × × × ×
× × × × ×
Operationalize machine learning and generative AI solutions Trainer Preparation Guide
Course Description
AI-300: Operationalize Machine Learning and Generative AI Solutions
Take ML models and AI agents from notebook to production. This 4-day hands-on Azure course covers the full operations lifecycle: MLOps with Azure Machine Learning and GitHub Actions, and GenAIOps with Microsoft Foundry. It suits beginners and seasoned engineers, and it prepares you for the AI-300 exam.
Duration: 4 days (about 28-32 hrs) | Format: Instructor-led or self-paced | Level: Beginner to advanced | Prerequisites: Basic Python and Git
What You’ll Learn
- Train and evaluate classification models with Azure Machine Learning
- Track experiments with MLflow, tune hyperparameters, and build ML pipelines
- Automate training and deployment with GitHub Actions
- Deploy and monitor models on managed online endpoints
- Set up Microsoft Foundry and build versioned AI agents
- Evaluate agents with cloud evaluators and optimize prompts
- Monitor and trace agents with Application Insights
- Improve agents with fine-tuning (SFT, RFT, DPO)
Who Is This Course For?
MLOps engineers, ML engineers, data scientists, DevOps engineers, and AI developers. Beginners get on-ramp content and guided labs. Experienced engineers can skip ahead to advanced topics.
Course Curriculum
-
ML lifecycle
-
Azure essentials
-
MLOps concepts
-
environment setup
-
Azure ML workspace
-
compute, data assets
-
training and evaluation metrics (accuracy precision, recall, AUC
-
Move from notebooks to scripts
-
MLflow tracking, hyperparameter tuning with sweep jobs
-
reusable ML pipelines
-
Source control for ML
-
CI/CD workflows
-
environments and secrets
-
triggers
-
model registration
-
Managed online endpoints
-
blue/green deployment
-
scoring, data drift
-
performance monitoring
-
retraining triggers
-
Projects
-
resources
-
model deployments
-
access control
-
GenAIOps lifecycle
-
Prompt engineering
-
agent instructions
-
tools
-
versioning
-
deployment
-
Test datasets
-
cloud evaluators (quality, safety, groundedness)
-
comparing prompt versions
-
evaluation gates in CI/CD.
-
Application Insights
-
distributed tracing
-
token and cost tracking
-
latency
-
debugging failures in production
-
When to fine-tune vs. prompt
-
supervised fine-tuning (SFT)
-
reinforcement fine-tuning (RFT)
-
direct preference optimization (DPO)
-
Build an automated pipeline that trains and deploys a model
-
then deploys a Foundry agent with evaluation gates and tracing
Course Eligibility
Audience profile
This course is intended for data scientists, MLOps dngineers and AI developers who design and implement
AI solutions using Azure Machine Learning and Microsoft Foundry. Students should be comfortable
working with Python, cloud services, and GitHub. Typical candidates are building or maintaining machine
learning pipelines, automating AI workflows, or operationalizing generative AI applications in enterprise
environments.
Audience prerequisites
Before attending this course, students should have: familiarity with Python programming; a basic
understanding of machine learning concepts; experience with Azure services (Azure portal, subscriptions,
resource groups); and familiarity with GitHub and version control. Experience with Azure Machine
Learning or Azure AI services is beneficial but not required. [NOTE: Confirm prerequisites from ADO
record.
Reviews
