Welcome to My Blog
Welcome to My Blog
I built this site to put my work in front of people — to show what I’ve built, what I know, and what I can offer the developer community and businesses. This is the first post.
I want to kick things off with something I’ve been living in lately: AI agentic engineering, and what it’s actually like from where I sit.
The productivity gains are real. I use AI in most of my work now and I’ve been able to build things I wouldn’t have gotten to otherwise. When you’re working solo or in a small team, having an AI that can generate a working implementation from a plain English description is powerful. You’re acting like a project manager — describing what you need, reviewing what comes back, directing the next step. The output can be built like a software engineer, by a software engineer. That part is genuinely useful.
But the barrier to entry dropping this low has a cost.
People are shipping applications at a business scale without the background to know what they’re actually shipping. The code is sloppy. The architecture ignores basic principles. Security vulnerabilities are baked in from the start because nobody reviewed them — or knew to. The tools make it feel like you’ve built something solid when you haven’t. We’re going to be dealing with the fallout from that for a while.
The AI takes shortcuts. If you’re not watching for it, it’ll duplicate code, build monolithic components that do everything, sprawl a single file across hundreds of lines when it should be broken into focused, single-purpose pieces. It reaches for whatever gets to passing output the fastest.
My job is to catch that and push back. When I see duplication, I have it clean up and go DRY. When something violates the Single Responsibility Principle — one class handling everything — I break it apart into proper utility functions. Large files get decomposed. That’s not just good practice; it actually works better with AI. Smaller, focused files consume less context. The AI gives you better results when it’s not trying to hold a thousand lines in its head at once. I work in monorepos with the backend and client in the same codebase, and I’ve leaned even harder into small, focused files since doing more AI-assisted development.
There’s something else worth naming, because I think it’s real: the fatigue. I’ve read the articles about developer burnout in the AI era and I get it now in a way I didn’t before. Managing that volume of code — reviewing it, testing it, catching what the AI missed — adds up in a way that’s hard to explain until you’ve been in it. My memory and mental load have shifted.
What’s strange is that AI development decouples your output from your cognitive state. You can be exhausted and still ship working code. That sounds like a win. But the question isn’t whether the code runs — it’s whether it has the features you actually need, whether the structure holds up over time, whether it’s secure, whether it handles edge cases. That part comes from testing and careful review, and it still takes a sharp mind. You can pump out a lot of code when you’re tired. Whether that code is any good is a different question.
I also write on Patreon and Ko-Fi — give me a follow there if you want to keep up. If you want to support the work, both platforms make that easy.