A technical deep dive into how we added a daily trader P&L section to the Briefing Hub, integrating paper trading audit logs with Kalshi's public API, while enforcing strict safety rules and COI blocklists.
How jTrader's exit system evolved from a dumb timer to a dual-threshold edge detector — and why 'never guess' is the hardest design principle to follow.
A Python regex analyst missed GitHub Actions workflows because of a missing '/' prefix in a string comparison. Here's how a one-character typo shipped to production and how adding CI/CD to the tool itself finally caught it.
How I learned Go and Rust by building personal AI tools, made a lot of mistakes, and somehow ended up with working software.
How joshbot's provider reconfiguration fix evolved through three commits to properly update agents.defaults.model in Go configuration.
Adding GitHub Copilot OAuth to joshbot took twenty minutes to implement and one very long day to actually make work. A story about hidden state, path bugs, and the debug log that finally explained everything.
From chicken mobs to chopping vegetables at alarming speeds - my journey into making games that probably shouldn't exist.
A canvas game, five production bugs, and what each one reveals about why certain failures are invisible to the person who wrote the code.
How a simple 'center the game' request turned into a complete rewrite, a canvas rendering bug I'd missed for months, and a lesson in agentic debugging.
What happens when you hand an agentic coding tool a real production codebase, say 'go until completion,' and actually mean it.
The chaotic, occasionally painful, and ultimately rewarding journey of building an AI-powered SDLC framework for regulated environments.
Why I'm moving my AI workflow to the terminal with OpenCode, a privacy-first, open-source coding agent.
Rebuilding this space to share what I've learned about software engineering, system architecture, and leading with empathy.