Personally, I think much of what makes the best explanations great is not the writing as much as charts and illustrations and other supplementary materials that make it easier to understand how things go together. While not technical or a blog, I like The Book of Shiatsu by Paul Lundberg as a great example of this due to the art style really helping with the explanation (you can get an idea from the Amazon preview (towards the end)[0]). I've seen one or two CS blogs that are good like that too but I don't have any links handy. Exactly how that would look depends on what you would like to create, I'm just suggesting that giving some thought to the artistic style of presentation can really draw people in and clarify things. Even if you aren't an artist, picking what to put in charts and tables or what code snippets to include can make a huge difference in presentation and even minor stylistic touches can make your explanation easier to remember (also complete and accurate explanations of the supplementary materials are extremely valuable and often overlooked even when a helpful illustration is given). Even things like taking pictures of your notes on paper or screenshots of whatever tools you use can add a more personal touch and give your readers something not found in the textbook (sometimes really mundane stuff turns out to be the most helpful). Or since you are interested in machine learning, maybe put that to work to create a unique style. Of course, you also need to decide how much time to spend on it and I don't mean figure out the best thing right away, just give thought to things other than text as you go. Some dramatic text styles can be helpful and fun too (like up goerizing[1]).

Challenging yourself to explain things quite a bit differenty than the materials you are learning from is likely to be a better experience for both you and your reader (who may well be reading the same texbooks as you anyway).

[0] https://www.amazon.com/Book-Shiatsu-Complete-Pressure-Manipu...

[1] https://xkcd.com/1133/