Linear algebra is the quiet engine behind machine learning, computer graphics, quantum mechanics, and statistics. But it is famously easy to "pass" and never understand, because it can be taught as pure matrix bookkeeping. Read it in the right order and the geometric meaning stays in front of you the whole time.
The core tension is between the abstract, vector-space view (clean, powerful, a little cold) and the concrete, matrix-and-application view (intuitive, but easy to mistake for the whole subject). The path below deliberately reads both so neither one calcifies into your only picture.
Two complementary starts
Begin with Linear Algebra Done Right by Sheldon Jay Axler, which builds the theory from vector spaces and deliberately delays determinants so you learn what the objects actually are. Read it alongside Linear algebra and its applications by Gilbert Strang, whose famously concrete, geometry-first approach is the perfect counterweight. Between these two you get both souls of the subject.
If Strang's applications text feels dense as a first pass, Introduction to linear algebra, also by Strang, is the gentler on-ramp built around his renowned lectures.
Deepen the theory
Once the basics are solid, Matrix Analysis by Roger A. Horn is the reference that treats matrices as serious mathematical objects, with the eigenvalue and factorization theory you will keep returning to. For a rigorous, elegant, purely theoretical treatment, Linear Algebra by Georgi E. Shilov gives the classical Russian-school view and rewards careful reading.
Where it actually gets used
Theory is only half the job; real matrices are enormous and imperfect. Numerical linear algebra by Lloyd N. Trefethen is the beautiful, readable introduction to how these computations are actually done on a computer, why some methods are stable and others quietly fall apart, which is essential if you will ever run this math at scale.
Finally, if your goal is data science or AI, Mathematics for Machine Learning by Marc Peter Deisenroth ties the whole subject to the applications that probably brought you here, showing exactly which pieces of linear algebra power modern models.
Read these in order and you will hold the geometry, the abstract structure, and the computational reality all at once, which is what it means to actually know linear algebra. Follow the full path to build that complete picture.
