Machine Learning Operations Specialization (Coursera)

MLops by Duke University

What you'll learn:

• Master Python fundamentals, MLOps principles, and data management to build and deploy ML models in production environments.

• Utilize Amazon Sagemaker / AWS, Azure, MLflow, and Hugging Face for end-to-end ML solutions, pipeline creation, and API development.

• Fine-tune and deploy Large Language Models (LLMs) and containerized models using the ONNX format with Hugging Face.

• Design a full MLOps pipeline with MLflow, managing projects, models, and tracking system features.

Specialization - 4 course series

This comprehensive course series is perfect for individuals with programming knowledge such as software developers, data scientists, and researchers. You'll acquire critical MLOps skills, including the use of Python and Rust, utilizing GitHub Copilot to enhance productivity, and leveraging platforms like Amazon SageMaker, Azure ML, and MLflow. You'll also learn how to fine-tune Large Language Models (LLMs) using Hugging Face and understand the deployment of sustainable and efficient binary embedded models in the ONNX format, setting you up for success in the ever-evolving field of MLOps

Through this series, you will begin to learn skills for various career paths:

1. Data Science - Analyze and interpret complex data sets, develop ML models, implement data management, and drive data-driven decision making.

2. Machine Learning Engineering - Design, build, and deploy ML models and systems to solve real-world problems.

3. Cloud ML Solutions Architect - Leverage cloud platforms like AWS and Azure to architect and manage ML solutions in a scalable, cost-effective manner.

4. Artificial Intelligence (AI) Product Management - Bridge the gap between business, engineering, and data science teams to deliver impactful AI/ML products.

Applied Learning Project

Explore and practice your MLOps skills with hands-on practice exercises and Github repositories.

1. Building a Python script to automate data preprocessing and feature extraction for machine learning models.

2. Developing a real-world ML/AI solution using AI pair programming and GitHub Copilot, showcasing your ability to collaborate with AI.

3. Creating web applications and command-line tools for ML model interaction using Gradio, Hugging Face, and the Click framework.

4. Implementing GPU-accelerated ML tasks using Rust for improved performance and efficiency.

5. Training, optimizing, and deploying ML models on Amazon SageMaker and Azure ML for cloud-based MLOps.

6. Designing a full MLOps pipeline with MLflow, managing projects, models, and tracking system features.

7. Fine-tuning and deploying Large Language Models (LLMs) and containerized models using the ONNX format with Hugging Face. Creating interactive demos to effectively showcase your work and advancements.

COURSE 1

Python Essentials for MLOps

• Work with logic in Python, assigning variables and using different data structures.

• Write, run and debug tests using Pytest to validate your work.

• Interact with APIs and SDKs to build command-line tools and HTTP APIs to solve and automate Machine Learning problems.

COURSE 2

DevOps, DataOps, MLOps

• Build operations pipelines using DevOps, DataOps, and MLOps

• Explain the principles and practices of MLOps (i.e., data management, model training and development, continuous integration and delivery, etc.)

• Build and deploy machine learning models in a production environment using MLOps tools and platforms.

COURSE 3

MLOps Platforms: Amazon SageMaker and Azure ML

• Apply exploratory data analysis (EDA) techniques to data science problems and datasets.

• Build machine learning modeling solutions using both AWS and Azure technology.

• Train and deploy machine learning solutions to a production environment using cloud technology.

COURSE 4

MLOps Tools: MLflow and Hugging Face

• Create new MLflow projects to create and register models.

• Use Hugging Face models and datasets to build your own APIs.

• Package and deploy Hugging Face to the Cloud using automation.