I’m excited to be hacking on something again.

A few days ago I heard the OpenClaw dude use the phrase “agentic engineering.” It stuck with me and comforted me. Despite having worked as an engineer my whole life (and even surviving 6 years at Meta), I still feel imposter syndrome. And even more so with the AI era. I don’t write code anymore. Heck, I don’t really read code either. So the question creeps in: am I still doing software “engineering?”

I got into this field because it let me create. And right now, I’m creating like never before. Everyday I do something new that I thought was impossible yesterday. In 2018, I dreamed of building a model to detect which way shoes were facing and using that to normalize images on my sneaker shopping platform. But it was just a dream. I had no idea where to begin. I tried taking a computer vision class, but got lost in the math. But now in 2026, it took a few hours to prototype and a few days to get reliable in production. I’m doing stuff I only dreamed of before. I’m renting GPUs by the minute. Parsing and normalizing thousands of footwear size formats. Testing multiple approaches to any problem with solid experiment results in hours.

Before AI, all of this required a team. Now I can do it alone.

This feels like engineering.

When I think back to my Facebook days, I remember beautiful systems. And a ton of terrible code. There were incredible engineers, yes. But there were also rushed reviews, rubber-stamped diffs (err PRs), and bugs everywhere. Some code barely worked. Some didn’t work at all. And yet the product shipped. The company still printed money.

Software engineering has never been about writing pristine code in isolation. It’s about shipping systems that work.

I often hear people say, “If I use AI, I at least read every line it writes.” That sounds like a nice principle. But how much more will you really catch? Even at Meta, engineers constantly worked with code they didn’t write and didn’t fully understand. I spent years learning to navigate unknown codebases, reverse-engineer intent, and ship changes safely. If I could do that inside millions of lines of legacy code, why can’t I do it with AI-generated code in my own app?

The engineering skill hasn’t disappeared. It’s shifted.

I define the architecture, train the model, benchmark performance, rent infrastructure strategically, deploy safely, and monitor outcomes. That’s engineering, right?

AI doesn’t remove engineering. It amplifies it.

I’m not less of an engineer because I don’t deal with code directly anymore. I’m more of one. I can now build things that were previously inaccessible to me.

I’m building. And that’s why I became an engineer in the first place.