Foundations of Open Generative AI Engineering

Overview

The Foundations of Open Generative AI Engineering course introduces learners to the principles, architectures, and trade-offs that define the open generative AI landscape. Starting with the distinctions between open source, open weights, and open access models, learners explore different licensing frameworks—including MIT, Apache, and CreativeML Open RAIL-M—and their implications for commercial use, attribution, and compliance. The course then covers the core architectures of open large language models (LLMs) such as Llama, Mistral, and Mixtral, alongside diffusion models used for image generation. Learners analyze how factors like parameter size, context windows, and inference speed impact performance and suitability for different applications. The final module develops a structured decision-making framework for evaluating open vs. closed models, balancing cost, scalability, customization, privacy, and data sovereignty. By completing a model selection analysis report, learners gain the ability to critically assess and recommend appropriate generative AI models for real-world use cases.

Syllabus

Foundations of Open Generative AI Engineering

Understand what sets open generative AI models apart. In this course, you’ll learn how to spot license limitations, break down the key components of large language and diffusion models, and compare performance trade-offs like parameter size, speed, and accuracy. You’ll also apply a simple decision framework that helps you choose the right model for your needs, building the confidence to make smart, compliant, and cost-effective choices.

Understanding Open Models and Licensing

In this module, you’ll learn how to tell what you can and can’t do with open generative AI models. We’ll break down the differences between open source, open weights, and open access, compare common license types, and show how each affects commercial use. You’ll also practice spotting legal red flags, understanding attribution requirements, and applying compliance best practices so you can avoid costly mistakes and deploy models with confidence.

Model Architectures and Capabilities

In this module, you’ll learn the fundamentals that shape how open models are built and perform. You’ll explore the core components of transformer architecture, compare major models like LLaMA and Mistral, and understand the principles behind diffusion models. You’ll also evaluate how parameter size, context windows, and inference speed trade off against each other, so you can make informed choices about which model fits your needs.