Learning in Public: An Architect’s Hands-On Journey

Most senior practitioners start the same way. Claude Chat. A few prompts. Genuinely impressive outputs. The instinct that this changes something significant, even if it is not yet clear what.

That is where I started. And for a while it felt like enough. I could use AI to think through architecture problems, draft documents, pressure-test ideas. Useful. Genuinely so. But useful is not the same as understanding. And understanding is not the same as knowing how to deploy this reliably at enterprise scale, in a regulated environment, with real governance consequences attached to every output.

I am a principal architect with twenty-five years in financial services. What I did not have, until recently, was the hands-on depth to design and build the AI integration layer myself. Not from a whitepaper. From actual builds that break in real ways.

So I started building. This section documents what that looks like in practice. Each article comes from something I actually built, a decision I actually faced, or a failure mode I actually observed. The writing follows the build, not the other way around.

I am not writing from the other side of the learning. I am writing from inside it.