A chronological build log of learning to ship AI-powered products in public. Not polished thought leadership — real stories: the setup, the tension, the breakthrough, and the reflection.
I volunteer on a sub-committee that shares job opportunities with members of Protocol Directorate. The old process worked — until it didn't. Building a better one taught me more about workflow than it did about code.
The expense tracker proved I could ship. But I needed somewhere to put it. Building mekanova.dev meant figuring out not just how to code a portfolio — but what story it should tell.
One project shipped, one journey documented. Before starting the next one, I wanted to write down what actually changed — in my skills, my thinking, and my sense of what's possible.
The app worked perfectly on my machine. Deploying it to Vercel took three attempts, two ESLint errors I'd never seen before, and a branch confusion I should have caught earlier. Here's the actual story.
The Vanderbilt course gave us a starting point. What I did with it — the look, the features, the three rounds of iteration — that part was mine. Here's how a practical assignment became something I'm actually proud of.
I'd used AI chat tools before. This was different. For the first time, the AI wasn't just answering questions — it was working alongside me, making decisions, and explaining why.
I wasn't looking for another certificate. I was looking for a way to stop being the person in the room who talks about AI and start being the person who actually builds with it.