OtherSelf paced YouTube courses, individual series run a few hours each·Free on YouTube, companion book Grokking Machine Learning sold separately

Serrano Academy (Luis Serrano)

4.5

If equations make your eyes glaze over, Luis Serrano is the person who finally makes the concept click. Serrano Academy is a free collection of the clearest explanations of hard machine learning ideas I have come across, and it has quietly become the place people send friends who are stuck on the maths.

What We Liked

Completely free, with a large and growing library of videos on YouTube
Unmatched at building intuition for tricky topics like PCA, SVMs, attention and transformers
Explanations lean on pictures, analogies and worked examples rather than dense formulas
Taught by someone with a maths PhD and serious industry pedigree, so the simplicity is earned, not shallow

What Could Be Better

Intuition first by design, so you will not come away with production ready coding skills
Content is spread across playlists rather than one guided path, so you have to sequence it yourself
Breadth and clarity are the priority, meaning some topics stop just short of full depth
You really need to supplement it with hands on practice to make anything stick

Detailed review

There is a particular kind of teacher who can take something that looks terrifying on a whiteboard and make you wonder why you were ever scared of it, and Luis Serrano is one of them. His academy is free, lives mostly on YouTube, and covers a wide arc from the fundamentals of machine learning through the underlying maths and, increasingly, the workings of large language models and transformers. Serrano's background matters here. He has a doctorate in mathematics and has worked as an AI educator and researcher at places like Google, Apple, Udacity and Cohere, so when he chooses to skip the formalism it is a deliberate teaching decision rather than an inability to go deeper.

The result is explanations that use cartoons, analogies and concrete examples to get the idea across first, then bring in the notation only once you already understand what it is describing. For topics that reliably scare newcomers, principal component analysis, support vector machines, the attention mechanism, how a transformer actually routes information, this approach is close to the gold standard. The companion book, Grokking Machine Learning, carries the same friendly, jargon light style into print and only assumes basic algebra, so the two pair naturally. The honest limitation is the flip side of the strength.

Because the emphasis is on understanding rather than building, you will not leave with a production workflow or a portfolio of code, and the material is organised as a set of playlists rather than a single structured curriculum with checkpoints, so you have to impose your own order and your own practice. Clarity also occasionally comes at the cost of the last mile of depth, which is a fair trade for the intended audience but worth knowing. My honest take is that this is the resource you reach for when a concept refuses to click, and one of the best free things on the internet for that job. Build your intuition here, then go and get your hands dirty somewhere more code focused, and you will find the practical courses far easier because you finally understand what the code is doing.

[ final ]

The verdict.

The best free starting point I know for anyone who understands machine learning ideas better through a good picture than a page of algebra. Use it to build genuine intuition, pair it with the Grokking Machine Learning book if you want the same style in print, and then take that understanding into a hands on course, because on its own it explains beautifully but does not put a keyboard under your fingers.