Why finance, and not something easier

On the desk

JavaScript and TypeScript, Python, Bash, React, Node, Postgres, Linux, Docker, DNS, Postfix, serverless platforms, Git, and the usual LLM APIs.

I got here through problems rather than coursework. I wanted mail on my own domain, so I learned to run a mail server properly. A camp needed to know where its groups were, so I built the tracker. A friend and I watched manpower scheduling happen entirely in group chats, so we started Jobxy.

Financial systems are where I want to keep working, because they demand what I already like in software: correct, auditable, and boring under pressure. A payment that is mostly right is wrong. That constraint suits me better than a looser one would.

AI is the other half. I spend real time on prompt design and evaluation, because getting a model to behave the same way on the hundredth run as the first is closer to interface work than to magic.

What I am after: an internship where I own something small end to end, and get told plainly when I am wrong.

Reading and building now

Payments and settlement — the genuinely hard part of fintech.
Getting Jobxy from a working prototype to something people pay for.
Writing evaluations for prompt systems instead of eyeballing output.