Data Engineer looking to explore ML and also transition

Working as a Data Engineer I feel I can approach ML but feel overwhelmed looking at current ML advancements. I remember people using MNIST dataset, and IRIS dataset and experimenting around and now I see LLMs, VectorDBs, etc. I can read and watch tutorials on YT all I want, but is it close to the real-world work being done? I hardly see anyone mention what goes on in their organizations with ML or MLOPS. I don't want to be in the backseat when things move faster than today.