← All Side Quests

What's a Side Quest?

Completed
What Why How

Why?

One of my trepidations about this wild new world of agentic software development / Vibe coding / Vibe engineering / context engineering… whatever you call it, is that it seems like all this tooling is just stunting growth. It takes longer to really learn a new language because you’re bashing your head against the wall less. But, on the flipside, the tools seem like they’re shrinking the internet somehow.

I still think creating projects is the best way to dive into a technology. Even if much of the code is written by the model. The new knowledge doesn’t imprint in the same way (you don’t really know the project quite as well), but I think the key to surviving in the new paradigm will be cracking the code on how to make it all stick.

So I keep tinkering on things that are bouncing around in my head.

What is a Side Quest?

The side-quests are just random projects that I have been thinking about for ages but I’m always too busy working, building my house, riding my bike, hanging out with my beautiful wife (while building and riding bikes).

I try to incorporate at least one new concept/thing/library/language/agentic tool into each side-quest.

Some current ideas / things that i’m intermittently working on:

  • TrainingLog - An iOS/Android app that can be used to track workouts. It will be customizable in that you can add your own sports, use whatever units you want on a sport-by-sport basis, send completed workouts to whatever server you want. Otherwise all the data live on your device.

  • Training-Server - This is just a simple rails webapp that can run locally on a Raspberry Pi that i’m using to store my own training data. Ultimately i’d like to be able to view heatmaps, training data, all kinds of metrics, and maybe incorporate some modeling into this.

  • Real TSS - I’m sitting on 10 years of detailed training data from Strava, and I have a hunch that given the limited accuracy of a lot of my biometric data (wrist based HR… come on), it might be possible to get a better picture of how “ready” you are to do some hard workouts by using the title/description of your workouts over time. I think using some old-fashioned ML modeling coupled with sentiment scores based on what you write down for the title/description… This might actually be a pretty good combination for figuring out how you really feel. There seems to be some literature precedent for using sentiment to score readiness/training stress. I just know personally, when i start to feel tired and down, my titles and descriptions get less creative. Or they just get really intense and brooding.