I learned from a prospective employer that the work I and other employees have done is more that of “power users.” It wasn’t diminishing what I’ve done, just a more accurate description. Still made me want to level up.
So one direction is either getting the best out of my experience as a power user or going beyond prompting and response evaluation. Now I’m in the learning trenches, battling it out with my lack of knowledge. I’ve been going deeper into the world behind the tools we use every day.
I’m learning the difference between a few extended career paths, and the first set of options are: LLM evaluation engineers, machine learning research engineers, and applied AI scientists. How do they think, how do they test, how do they iterate, and how do they measure performance beyond just “good output”?
Understanding prompt intuition is only one layer. (It’s different from my JavaScript skills as well, though those help.) Evaluation design, controlled experimentation, model behavior analysis, and measurable improvement are what separate power users from builders. I’m putting on my lab coat.
I’ll keep you updated on how this goes.