I think self-studying a university curriculum is also helpful. I've studied physics at the undergraduate level myself.
Quantum mechanics, for example, is useful enough as a general background, and going through fluid dynamics and GPS problems is also helpful. But if you just follow the application problems, you can learn what you need, but the bigger risk is that you might miss the larger framework that defines those problems. That's not necessarily a bad thing. In fact, the entire university curriculum training is ultimately a process of translating complex phenomena into the Western scientific way of understanding things. It's a mental model that says 'this phenomenon can be interpreted with this kind of formula.'
In other words, it's about building mental models. In a formal university curriculum, you usually learn things like vectors, topological spaces, energy conservation, and how to map real-world phenomena onto these perspectives. It's about learning to simplify the world using mathematical tools. I didn't go to a top university, so I didn't learn things like tensors, but I hear they're taught now. I used tensors in grad school.
Of course, when I actually code and deliver factory equipment, I've done motor-related work under NDA, and the actual formulas aren't always perfectly accurate. There are corrections and adjustments. But the important thing is not just problem-solving itself, but building the mental model of 'how to approach the problem' before solving it. And I think the curriculum helps with that.