How to Learn AI from Scratch for Free
A roadmap to learn AI fundamentals for free including prerequisites, AI engineering, ML engineering, and more
I set up a Machine Learning Roadmap for the AI for Software Engineers community a few years ago to share high-quality, free machine learning learning resources in order of how to learn them. The roadmap takes anyone from wherever they are in their CS/AI journey to understanding AI from fundamentals.
Today, I’ve just finished the first major revision to make the roadmap even better. Please support it by adding a start on GitHub.
AI and ML engineering have been more explicitly added
These topics have their own sections with their own resources instead of being included in the machine learning topics section. There are enough important resources for each and each role is well enough defined in industry to warrant sections of their own.
There’s now also the option for software engineers to go straight from prerequisites to AI engineering without needing to get too deep into ML fundamentals. I highly suggest going through ML fundamentals anyway (or going back to them after finishing the AI engineering section) as understanding the fundamentals of AI will pay dividends in the long-term.
Duplicate topics removed to streamline the roadmap further
I’ve removed resources that didn’t prove useful and added more resources where there were gaps. Specifically, the added topics are LLMs, AI engineering, and ML engineering. I found them to be particularly weak. I’ve also removed duplicate topics already covered by the Google Machine Learning Crash Course.
As mentioned above, I’ve added a streamlined AI engineering roadmap for engineers wanting to onboard to building with AI faster.
Supplemental paid resources added
I’ve added supplemental paid resources. The focus of the roadmap is still on free resources and the entire roadmap is free. I’ve added paid resources that further streamline the learning for those who want to purchase them. The sections where this is the case have been annotated with a paid resource block quote (see below).
All paid resources are resources I highly recommend either because I’ve read them myself or trust the educator/author behind them. Paid resources are from the top AI educators in the world and will always be optional and properly vetted.
Combined the AI for SWEs repo with the ML roadmap
I realized the hands-on resources I’ve created for the newsletter repo fit into the roadmap. The roadmap is also a much better resource for learning. Instead of finding random topical hands-on exercises in a standalone repo, readers can instead consult the roadmap for a much more organized learning experience.
Thus, the AI for Software Engineers repo is now combined with the ML roadmap repo and I’ll be continually adding resources as I find and create them. The old repo has a notice in the README to redirect visitors.
You can now contribute for swag
The ML roadmap now takes contributions! I’d love for this to be a crowdsourced effort to make the most straightforward and complete learning resource for AI fundamentals. High-quality, original contributions readers have created are encouraged. I’ll review everything submitted and maintain a high bar to ensure roadmap quality. See the contribution guide for more information.
If you contribute an original guide (and you’re in the US), I’ll send you a piece of AI for Software Engineers swag. I’ll be setting up a merchandise store soon and it’ll come directly from that. If you add something in the near future, it might be a bit before the store is fully setup and I can send something out.
[Beta] Terminal agents have been added to supplement your learning
I’ve added instructions for CLI terminal agents to help walk you through the guide. This is experimental and still in testing, but I’m hoping these agents can supplement the resources and better personalize the roadmap for each reader.
To try this, fork or clone the ML roadmap repo and start your favorite terminal agent within the directory. This will be improved over time to further personalize the learning experience.
Enjoy the roadmap! Feedback is always encouraged. Feel free to submit a PR according to the guidelines to contribute. Don’t forget to star the roadmap.
Always be (machine) learning,
Logan