a set of small, focused projects exploring systems, infrastructure, browser tooling, data modeling, and agent-native interfaces.
each folder is a self-contained build answering one technical question end-to-end.
every experiment:
- proves a single capability (scraping, realtime events, browser interception, agent apis, etc.)
- defines a clear contract (events, schema, api, or interface)
- includes a demo and reproducible setup
- documents real engineering learnings
each folder contains:
README.md → goal, demo, architecture, learnings
src/ → minimal implementation
demo/ → gifs or screenshots
schema|api/ → interface or data contract
reading the README alone should explain:
- the problem
- the approach
- failures and fixes
- tradeoffs and next steps
to break down complex systems into small, testable builds and understand them deeply.
many experiments feed into larger projects over time.
current themes include:
- browser infrastructure
- agent-first apis
- realtime data pipelines
- scraping and document normalization
- structured knowledge systems
every experiment ends with:
- what broke and why
- what worked
- performance notes
- design tradeoffs
focus is on insight, not lines of code.
each project builds on the last. patterns get reused. systems get sharper.