Day 13 - Specs are worth the effort

Yesterday I gave myself a bit of a break and went to a used goods store to buy my family some matcha bowls and myself some yugioh cards. I worked in the morning on mapping out my v0.1 spec for my knowledge graph as a recommender system + defining the tests I wanted to run.

After the thrift store detour, I came back home to eat with the expectation that I would present at an engineering event in the evening. Turns out I had not signed up using the right link so I was not given a chance to present - sad.

I decided to avoid going to the event because it would have taken me 1 hour to get there and my main goal right now is to focus on building my app or speaking to people who might help me build my app.

So I spent the afternoon and evening working on my spec. Once I had a defined spec, I fed it both to Claude and Gemini. For Claude, I use general Sonnet - since Opus or 'deep thinking' options are somehow blocked for me with a large context window. For gemini, I use the Deep Research when I want it to review a spec.

And I spent the rest of the evening reviewing the feedback from both models, thinking about what I liked, having detour discussions about relevant topics (e.g., tenant segregation in the knowledge graph, data privacy concerns). And at the end of the day I was left with a better spec but I want to go through the process again this morning -> sonnet + gemini deep research.

I need to share my progress with using a vision model (movenet) with a colleague I met here so I need to shift focus a bit after I'm done with my spec. Then gym. And I want to practice judo while in Japan but before enrolling at the Kodokan, I need to be interviewed by the instructor.