Then there's the "culture" of development. One day you're implementing hexagonal architecture to feel sophisticated, and the next you're arguing about whether extending base classes is a sin or a virtue depending on which community you're annoying that week.
Add AI to the mix and the goalposts just sprint away from you. It’s not enough to use a wrapper; now you're supposed to understand the linear algebra behind the weights while simultaneously wrestling with the absolute disaster that is real-world data pipelines.
My current AI workflow for surviving this madness:
-
Offload the boilerplate: I use LLM agents to handle the "industry standard" CI/CD configs and testing suites that I'm too bored to write manually.
-
Rapid Prototyping: When my "entrepreneurial ambitions" kick in, I use a prompt engineering loop to build MVPs in hours rather than weeks.
-
The Knowledge Gap: When I hit a math concept in a paper that looks like ancient hieroglyphics, I feed the snippet to Claude and tell it to explain it to me like I'm a tired dev who just wants to go to sleep.
The list of things to know is dizzying. Marketing, finance, physics, Japanese culture (thanks, anime), and the general art of not being a robot. It's a losing game, but honestly, the chaos is the only part that keeps it interesting. The older I get, the faster the tech moves, and the more I realize that "mastery" is just a polite word for "I've googled this specific error five times today."
Next Multi-Voice Audiobooks: Solving Character Attribution →