Overview

This course features Coursera Coach! A smarter way to learn with interactive, real-time conversations that help you test your knowledge, challenge assumptions, and deepen your understanding as you progress through the course. Master the entire lifecycle of building and deploying machine learning systems in production with this hands-on DevOps to MLOps Bootcamp. You'll learn how MLOps optimizes model development, deployment, and monitoring, gaining skills in tools like Docker, Kubernetes, MLflow, FastAPI, Streamlit, Prometheus, and GitHub Actions. This course bridges the gap between data science and scalable ML infrastructure. You'll explore MLOps concepts, trace its evolution through LLMOps and AgenticAIOps, and study real-world case studies. Apply these principles through a regression-based house price prediction project. The course covers CI pipelines with GitHub Actions and advanced production systems with Kubernetes, KEDA, and ArgoCD. The final sections focus on monitoring, autoscaling, and implementing GitOps pipelines for ML/LLM app deployment. Ideal for data scientists, ML engineers, DevOps pros, and developers, the course requires basic Python, ML knowledge, and container familiarity. By the end, you'll deploy models with containerized APIs and manage scalable systems.

Syllabus

Introduction to MLOps

In this module, you will be introduced to MLOps, its core principles, and its importance in modern machine learning workflows. The evolution from traditional MLOps to emerging paradigms like LLMOps and AgenticAIOps will be covered. You'll also compare DevOps and MLOps, examining their similarities and differences, and explore the growing role of the MLOps Engineer.

Getting Started with the Use Case and Environment Setup

In this module, you will set up the environment and tools necessary to work on the house price prediction project. You'll get hands-on experience in setting up Docker containers, configuring MLflow for experiment tracking, and creating isolated Python virtual environments for reproducibility. Additionally, you'll understand the end-to-end ML lifecycle and how MLOps practices integrate into it.

From Raw Data to Models

This module focuses on preparing and transforming raw data for modeling. You will learn essential data engineering and feature engineering techniques, including how to split data for training and testing. Additionally, you will experiment with different algorithms and hyperparameter tuning to identify the optimal model configuration.

Packaging Model along with FastAPI Wrapper and Streamlit with Containers

In this module, you’ll transition from model development to deployment. You’ll learn to package your model with FastAPI and create a user interface with Streamlit. The module focuses on containerizing the application with Docker and Docker Compose to ensure the deployment is scalable and production-ready.

Setting up MLOps CI Workflow with GitHub Actions

This module covers the automation of MLOps pipelines using GitHub Actions for continuous integration (CI). You’ll learn to create workflows that automate the model training, testing, and deployment processes. The integration of MLflow and Docker will streamline model tracking and container management as part of the CI pipeline.

Building Scalable Prod Inference Infrastructure with Kubernetes

This module introduces Kubernetes as a platform for deploying scalable machine learning models in production. You will learn how to architect and deploy ML model serving infrastructure using Kubernetes, including configuring pods, services, and deployments. You'll also generate and customize Kubernetes YAML manifests to automate deployment and scaling.

Monitoring and Autoscaling an ML Model

In this module, you will focus on monitoring and autoscaling of machine learning models in production. Using Prometheus and Grafana, you'll implement system monitoring and visualize performance metrics. You'll also learn to automate scaling using KEDA and VPA based on resource usage, and conduct load testing to evaluate system capacity under stress.

GitOps Based Deployments for ML/LLM Apps

This module introduces GitOps principles and how they can streamline deployment in MLOps. You will learn how to use ArgoCD to implement continuous delivery (CD) pipelines and manage ML/LLM application deployments. By designing end-to-end CI/CD workflows, you’ll understand how GitOps ensures a seamless, automated deployment process for machine learning models.