Are you wanting to get into LLMs in particular or something else? I am a software engineer also trying to make headways into so-called "AI", but I have little interest in LLMs. For one, it's suffering from a major hype bubble right now. The second reason is that because of reason one, it has a huge amount of attention from people who study and work on this every day. It's not something I have the time commitment for to compete with that. Lastly, as mentioned, I have no interest in it and my understanding of them leads me to believe they have few interesting applications besides generating a huge amount of noise in society and dumping heat. The Internet, like blogs, articles, and even YouTube, are already being overrun by LLM-generated material that is effectively worthless. I'm not sure of the net positive for LLMs.
For me personally, I prefer to work backwards and then forwards. What I mean by that is that I want to understand the basics and fundamentals first. So, I'm, slowly, trying to bone up on my statistics, probability, and information theory and have targeted machine learning books that also take a fundamental approach. There's no end to books in this realm for neural networks, machine learning, etc., so it's hard to recommend beyond what I've just picked, and I'm just getting started anyway.
I wrote a couple of these books (and published them as open access)
Here's my one on computation probability. The code and math here underlie "AI". It's the same fundamentals, and even code libraries (Jax, pytorch etc(
https://bayesiancomputationbook.com/welcome.html
I also posted my more specific guidebook to the fundamentals of GenAI above. Hope both help
For me personally, I prefer to work backwards and then forwards. What I mean by that is that I want to understand the basics and fundamentals first. So, I'm, slowly, trying to bone up on my statistics, probability, and information theory and have targeted machine learning books that also take a fundamental approach. There's no end to books in this realm for neural networks, machine learning, etc., so it's hard to recommend beyond what I've just picked, and I'm just getting started anyway.
If you can get your employer to pay for it, MIT xPRO has courses on machine learning (https://xpro.mit.edu/programs/program-v1:xPRO+MLx/ and https://xpro.mit.edu/courses/course-v1:xPRO+GenAI/). These will likely give a pretty up to date overview of the technologies.