Writing on AI-augmented engineering, agent quality, and technical leadership. Summaries and SEO live here; full articles on Medium.
Before the story of scaling from ten orders to five thousand, here is what we actually run: what a commodity Murabaha is, why every step is legally load-bearing, and why stock has to exist as individually coded units.
In January 2025 a stress test of ten automated orders left four of them stuck. Eighteen months later the same platform ran 5,200 orders end to end at 99.94%. Ten case files on the code and database work that got it there.
A purchase step that took seven and a half seconds contained 521 milliseconds of actual work. The rest was queue wait. Five case files on the infrastructure half of the story: splitting queues, sizing a worker fleet, and the deploys that quietly skipped it.
It's 2 AM on a Tuesday, and your Slack is exploding. A customer just reported a critical bug: your AI agent is giving incorrect responses about payment processing. You fix it, run a few manual tests, and deploy — then unrelated features break by 6 AM.
