DURATION: 3 Day.
Note: Labs for almost all modules can be performed on open source. So Koenig DC can be provided. Student can also install software’s on their own system.
Module 1: MLOps Fundamentals
- Introduction to MLOps and its significance.
- Challenges in traditional ML model management.
- Solutions offered by MLOps.
Module 2: MLOps Toolbox
- Applying MLOps tools for end-to-end projects.
- Integration of tools: DVC, Git, MLFlow, and DagsHub.
Module 3: Model Versioning with MLFlow
- Versioning and registering ML models with MLFlow.
- MLlow’s role in managing ML lifecycle.
Module 4: Data Versioning with DVC
- Capturing data and model versions with DVC.
- On-premises and cloud storage integration.
Module 5: Creating Shared ML Repository
- Utilizing DagsHub, DVC, Git, and MLFlow for versioning.
- Collaborative ML model management.
Module 6: Auto-ML and Low-Code MLOps
- Automation of ML model development with Auto-ML and Pycaret.
- Streamlining model versioning, training, evaluation, and deployment.
Module 7: Explain ability and Auditability
- Understanding model interpretability and explain ability.
- Monitoring model performance and data drift with SHAP and Evidently.
Module 8: Containerized ML Workflow with Docker
- Packaging code and dependencies using Docker.
- Efficient distribution of Machine Learning applications.
Module 9: Deploying ML via APIs
- Model deployment through API development with FastAPI and Flask.
- Deploying APIs on Azure Cloud using containers.
Module 10: Deploying ML in Web Applications
- Developing web apps with embedded ML models using Gradio and Flask.
- Deploying to production in Azure via Docker containers.
Module 11: Automated ML Services with BentoML
- Introduction to BentoML and its role in automated ML service development.
- Putting BentoML services into production using Docker.
- Integration of BentoML with MLFlow.
Module 12: CI/CD with GitHub Actions and CML
- Introduction to GitHub Actions and Continuous Machine Learning (CML).
- Practical lab: GitHub Actions for MLOps CI/CD.
Module 13: Model Monitoring with Evidently A
- Monitoring models and services using Evidently AI.
- Identifying data drift and evaluating model quality.
Module 14: Model Monitoring with Deep checks
- Components of Deep checks: checks, conditions, and suites.
- Hands-on experience with Data Integrity Suite, Train Test Validation Suite, Model.
- Evaluation Suite, and Custom Performance Suite.
Module 15: Complete MLOps Project
- Developing an ML model from scratch.
- Validating code and pre-processing data.
- Versioning with MLFlow and DVC.
- Sharing repository with DagsHub and MLFlow.
- Building an API with BentoML.
- Creating a Streamlet app.
- Implementing CI/CD with GitHub Actions.
