In November 2022, ChatGPT completely blew my mind.
That moment pushed me down a rabbit hole. In 2023, I started reading about deep learning to understand how it works.
I picked up two books from one of my favorite tech publishers:
- How AI Works: From Sorcery to Science (2023) by Ronald T. Kneusel
- Deep Learning: A Visual Approach (2021) by Andrew Glassner
Then, in February 2025, Claude Code came out and somehow managed to blow my mind all over again. I liked it enough to subscribe to a pro plan.
I finally finished the second book, a massive work of over 700 pages. What makes it special is its heavy use of visuals. As a visual learner, that approach clicked with me.
Today, I'm publishing my reading notes as part of my living notes, which I'll continue improving as my understanding deepens.
Glassner generously released the book's illustrations under the MIT license. I took advantage of that:
- out of 800+ figures, I selected around 300
- many were redrawn as SVGs
- the notes are built around these visuals, closely following the structure of the book
Along the way, I also used:
- Anton Tsitsulin's 12-color qualitative (or categorical) palette
- the Roboto typeface
- ChatGPT and Sonnet to clarify and summarize parts of my notes, and to fix grammar (English isn't my first language)
It took much longer than I expected, but it was worth it.
These notes gave me a solid intuition, one I can build on should I decide to go deeper into the math or explore libraries like TensorFlow and PyTorch.
Joint