On prioritising what you want to learn
it's easy to get distracted by the current discourse on what people are learning (or posing to learn).
you need to be careful on prioritising what you want to learn.
somehow these days the meme has been let's do hardware, c is cracked, we will go cuda. it's downstream of what influential figures like karpathy are pursuing.
problem is, mere exposure can make you feel more interested in new things like gpu programming, writing llm in c and unconsciously prioritise the same when let's say you were already trying to learn about the transformer architecture. (shiny object syndrome sort of situation lol)
frequently hopping bandwagons will lead to scattered attention and you don't want that. now i get it, you gotta pursue your intellectual curiosity but you also need to think from your usecase point of view.
what do i require to learn more urgently? what will allow me to build the thing i want to build faster? what is more relevant for me based on my previous understanding? is it worth going this deep in the stack for now?
depending on the time you have, you can pursue more than 2 things in parallel but you gotta stick to what you have decided atleast for a few days.