Notes on engineering practice, workflow, and the small decisions that compound into how I ship software.
Model Context Protocol went from Anthropic side project to industry standard in under a year. Here's why it matters, how it works, and what it means for how we build AI-powered tools.
You don't have a QA team. You ARE the QA team. Here's how I decide what's worth writing Playwright tests for — and what I deliberately skip.
I SSH into servers, push code, and monitor production from the same device I use to doomscroll. My first dev machine couldn't even run Chrome.
I replaced a fragile SSH-and-pray deploy workflow with Dokploy, an open-source self-hosted platform. One install, Dockerfile deploys, automatic SSL, and a dashboard for $0/month. Here's the setup and what I learned.
I hooked up an AI agent to my terminal, file system, and messaging apps. It deploys my code, schedules my tasks, and remembers what I told it last month. Here's what actually happened.
Every project tempts you to try the new thing. Here's why I keep reaching for Postgres, Next.js, and a Dockerfile instead — and when boring is the wrong call.
Most Git advice is written for teams. Here's the workflow I actually use when I'm the only person committing — branches, commits, releases — without the team-of-twelve ceremony.
