It does not take very long after attending multiple courses on Claude Code for me to have the courage to start building some seriously useful apps. But I had to deal with some issues, like the ridiculously high rate at which I'm burning AI tokens, so I turned on OpenAI's ChatGPT to advise on how I can reduce the tokens I'm burning on Claude, and it recommended installing a few skills to lower my usage. In the process, I somehow learnt what a linter is, and I was marvelling at how I could be so far behind in software engineering yet still churning out new Windows programmes at such a rate.
So, armed with a somewhat lower burn rate, I proceeded to create a portfolio tracker that consolidates all my investments in one place (sans the portfolio picks my students make, which I track more religiously) and provides a helicopter view of my investments. So I wrote something that lets me import a CSV file from Yahoo Finance and display all my portfolio positions on one page.
Immediately, I have to confront the fact that, thanks to the semiconductor rebound, I have too much UMS across all my portfolios. (Believe me, it's a happy problem if you know how UMS has been doing lately)
Then I realised that I'm actually teaching a data analytics class at the Poly, and the same metrics I track are the ones I use, so I started adding a simple dashboard to the programme.
I can't really describe my class's excitement in words when I showed them my programme and explained that they could create it in 2-3 days. I have never taught a subject like this before.
People want to understand the consequences of what they learn in school and why a simple grounding in numbers can make a big difference in their lives.
This weekend, I'm going to stop working on my Portfolio Tracker and start thinking very deeply about a program that can perform an X-ray on any business contract that arrives in my hands or create notes for a new legal judgment.
If you can dream it, it can happen.
No comments:
Post a Comment