I've been building software for over 15 years, and for the last five, I've been hooked on AI development. It started as a way to boost my own workflow with smarter testing, automated code reviews, and instant prototypes. Then the rabbit hole opened, and I went all in. Now I'm building 8-stage prompt frameworks, self-improving RL feedback loops, synthetic training data pipelines, and real-time streaming media servers for in-meeting AI avatars.
These days I treat AI like any other part of the stack. I understand how it works, stress-test it, and only keep what's useful. I squeeze most of this work into early mornings and late nights, between managing 3½-year-old twins. When your time's precious, you quickly learn the difference between what sounds cool and what actually works.