I read Daniel Roy Greenfeld’s post on how he found that coding a lot was key to improving his skills. It makes sense. Everything I’ve read so far and my previous experience at the metaskill of learning new things tells me that it is a good investment of time.

Just like you get good at writing by doing a lot of writing, on some level that is true for coding. (Of course, there are additional pieces to the puzzle: you have to develop some taste alongside the pure production side, you have to do some quality-control and refactor your code, and so on and so on.)

For me, this looks like the following:

  • coding at work during the week
  • smaller focused exercises from PythonMorsels, Exercism, LeetCode and AlgoExpert
  • code written while working my way through the fastai course; this will probably manifest as blog posts here as well, outlining some small project I completed along the way.
  • a bigger project, perhaps a package, that I’ll start building at some point. I have some ideas for things I want to implement. I’ll pick one soon. It’ll probably be related in some way to the fastai coding. I’m thinking right now of making a tool that allows you to download PDFs and use the pages of those PDFs as image files in computer vision problems; a data ingestion tool, in other words.
  • smaller scripts to solve daily problems in my digital life. I’ll store those on my GitHub somewhere and write up the design decisions around the more interesting ones here.

One thing I took note of was how Daniel mentioned that it made sense to specialise and focus on one language at a time, particularly in the early days. Rather than indulging my curiosity and doing 1001 things using Go or lisp or whatever, I will try to stick to Python at least until I feel more confident with it.