Reliability & testing

Make failures understandable—and change less frightening

These essays look at error models, automated testing, runtime assumptions, and the support structures that let engineers change systems with confidence.

The crash that wasn't Debugging a CI-only failure with AI and turning diagnostics into product capability. No time for testing? Why testing is part of delivery, not optional work after implementation. How to think about errors in your code A practical model for deciding what failure means across applications and services. A taxonomy of applications Scripts, apps, services, and jobs—and why their runtime shape changes engineering decisions.

Distributed systems

Understand why the machines need each other

Start with the economic motivation for distribution, then move into communication, resource naming, and the abstractions that make remote systems feel local.

Why distributed systems are useful The problems that justify distribution in time, space, or both. Virtual filesystem: fun and profit Using paths as a general interface to local and remote resources. libstyxe: Great Refactoring Redesigning a 9P protocol library for extensibility.

Building in public

Learn through the projects themselves

Side projects are where architectural ideas meet awkward environments, dependency constraints, users, packaging, and the occasional false crash.

Kinjo A terminal UI for discovering and acting on local DNS-SD and mDNS services. Debugging Kinjo's CI with AI A concrete account of human intuition, fast experiments, and durable tooling. libsolace: library philosophy Capturing repeatedly useful engineering primitives in a focused C++ library.