Basically the deal with Alexander is he was a VERY early adopter of computerization in architecture. He made some kinda... missteps? about systematization in his earliest writings. Then walked it all back / developed it with "A City is Not a Tree" talking about graphs and urban organization.
His arc from there was constantly deeper into trying to make sense of how information science techniques could be applied, most notably in A Pattern Language and the case study of its application, The Oregon Project.
Then, his "magnum opus" was a development on that where he tried to fill in the gaps that A Pattern Language left was A Timeless Way of Building.
If you don't want to get too deep, I'd stick with A Pattern Language.
"The Nature of Order" (four books!) is the most extensive - beautiful, comprehensive, a little overwhelming, he lays out a broad philosophy of how built up spaces can foster life. Lots of visual examples, but gets pretty conceptual and abstract too.
"The Timeless Way of Building" is a bit more focused, centers around the architectural concerns. More compact, more practical than the Nature of Order.
"Pattern Language" -- you'll see that one suggested a lot. He lays out the concepts of interacting patterns ("best practices") that act on a variety of scales, some more generic, some more abstract. It's less of a read than a reference, but a few chapters lay out the motivations.
"Notes on the Synthesis of Form" - the most far out one of the four suggestions, abstract, philosophical, a rarefied read. Dissects the essence of concepts, defines what "form" is -- fascinating read for sure, and not too long.