From the fairway to the terminal

I came to programming the way most people come to golf: convinced it would be straightforward, immediately humbled, and somehow unable to stop.

I'd always wanted to learn — not to become a developer, but to be able to build things myself. Automate the repetitive. Stop waiting on someone else to write a script. For years, the barrier was high enough that it stayed on the list. Then AI made coding accessible, and the excuse ran out.

Before writing a single line of code, I sat through Andrew Ng's machine learning course on Coursera. Early mornings, kids still asleep, headphones in. It was dense — gradient descent, neural network architecture, things I will never implement from scratch. But it gave me something I didn't expect: a mental model of what's actually happening inside these systems. The engine, before the steering wheel. Not strictly necessary, but it meant I could read between the lines later.

Hindsight makes this feel smarter than it was. At the time it mostly felt slow.

Then came the free-for-all: Harvard CS50, Kaggle, YouTube at 1.5x speed. I picked up Python. Started something small — a tool to automate my morning headline routine. Suddenly I needed to understand APIs, environment setup, version control. Each answer opened three more questions. It felt like a compacted CS degree, delivered in the wrong order.

What I've built so far is modest — a multi-agent system that brainstorms and critiques investment ideas, an RSS pipeline that summarises finance news so I don't have to. Neither will make the front page of Hacker News. Both solved real problems I had, and both taught me more than any tutorial.

Still building. Still Googling embarrassing things. Enjoying it more than I expected.