Carbon Aware Computing for GenAI Developers

What You'll Learn

• Retrieve real-time data on global energy mixes and carbon intensity from the ElectricityMaps API.
• Run machine learning training jobs using low-carbon electricity by selecting cloud server locations based on carbon intensity.
• Analyze the carbon footprint of Google Cloud usage data including machine learning training, inference, and storage.

About This Course

This course teaches GenAI developers how to reduce the environmental impact of their machine learning workflows by leveraging clean energy sources and optimizing job locations. Throughout the lessons, you’ll learn:

• Query real-time electricity grid data and analyze the power sources and carbon intensity of various regions.
• Select low-carbon regions for model training, using real-time grid data from ElectricityMaps.
• Retrieve carbon footprint measurements for ongoing cloud jobs using the Google Cloud Carbon Footprint tool.

By the end, you’ll be equipped to make environmentally conscious decisions when deploying machine learning workflows, using tools like ElectricityMaps and Google Cloud’s carbon tracking features.

Course Outline

• Introduction
  Overview of carbon-aware computing for GenAI developers.
• The Carbon Footprint of Machine Learning
  Exploring the environmental impact of machine learning and data processing.
• Exploring Carbon Intensity on the Grid
  Using ElectricityMaps to retrieve carbon intensity and energy source data.
• Training Models in Low Carbon Regions
  Selecting regions with low carbon intensity for deploying training jobs.
• Using Real-Time Energy Data for Low-Carbon Training
  Leveraging real-time grid data to optimize training locations.
• Understanding your Google Cloud Footprint
  Analyzing and tracking the carbon footprint of Google Cloud-based machine learning activities.
• Next Steps
  Practical steps to integrate carbon-aware computing into workflows.
• Conclusion
  Summary of key takeaways and future considerations.
• Google Cloud Setup
  Setting up and configuring Google Cloud tools for carbon tracking.

Who Should Join?

This course is ideal for developers with Python experience interested in reducing the carbon footprint of machine learning workflows.