Building unbuffered: week 0
This site exists to lower the cost of writing things down. The plan is four kinds of output, and the architecture had to serve all of them without four different workflows:
- Interview prep — algorithms, data structures, and system-design notes that get revisited and grow. These live in the garden, where a note has an explicit maturity stage and mindmaps are just Mermaid blocks.
- Engineering logs — what I tried and learnt this week. Long-form goes here in /blog; anything that fits in a terminal screen goes to /tips.
- Home experiments — AI, non-AI, and financial learnings. Write-ups become blog posts; reference topics become garden notes; anything that turns into a real system earns a project.
- Teaching the kids — coding, robotics, public speaking. That’s the kids lab, deliberately fenced off from the professional stream.
Everything is Markdown, validated against a typed schema at build time, deployed by
git push. The stack decision is recorded properly in
ADR-001; the pipeline has its own
project with measured numbers.
Week 0 lesson: the hardest part wasn’t the code — it was naming things.