AI Engineering

I’m about to start writing software mainly on top of foundation models.

Over the past four years, I have learned how to tell visual news stories mainly consumed on the vertical phone screen. I have managed to build full-stack apps that scale on-demand and propagate to many channels.

I still get a sense of exhilaration when I figure out the best way to arrange a certain set of building blocks to solve a particular set of problems.

I have used LLMs as coding helpers for the past year in my day job, quite regularly. They are getting better, and have been useful for this entire time.

In fact, multimodal AI models have gotten so good, and there are so many interesting possibilities to build on top of, that I feel genuinely excited about tech again!

My frustrations with AI remain, including the fact that I misclicked the previous sentence about “my frustrations” in VS Code when writing this very text, and Copilot started sending a prompt to a server to fix it.

I was going to tweak the wording, but I think I’m going now to just leave it as it is, as an act of human rebellion. Fast drafts, get initial ideas out of my head, clarify my own thinking, that’s at least part of why this blog exists.

I’ll see if I manage to become more active as a blog poster, too. I’ve written mostly fictional and creative stuff elsewhere lately, and I’m looking forward to sharing that as well when the time is right.

K