Learning
Learning tracks what I'm actively studying and in which direction I'm developing as an engineer — the process, not the finished result. Blog is where a topic becomes a synthesis once it's worked through; Projects is where it gets applied; Changelog is what changed in Developer OS itself. Learning is what's still in progress.
Currently learning
AI Engineering & Agent Orchestration
How to direct multiple AI agents toward a shared specification instead of letting any single one drive unsupervised.
Workflow design for multi-agent review — Claude implementing, Codex reviewing independently — and where human-in-the-loop checkpoints actually change the outcome versus where they just add ceremony.
Related article →Software Architecture & Decision Records
Writing architecture as explicit, falsifiable decisions instead of implicit tribal knowledge that only lives in someone’s head.
Structuring trade-offs as durable properties (ADRs) that survive a tool swap, and knowing when a decision deserves that weight versus when it’s just an implementation detail.
Related project →
Recently explored
Static-first content architecture
Content pipelines that validate and compile at build time, with no database and no runtime parsing — and where that model starts to strain.
Learning philosophy
Learning here isn't a course list or a progress bar. There's no percentage complete, no certificate, no streak to keep. A topic stays on this page for as long as it's actively shaping decisions in Developer OS, and moves to "recently explored" once it's settled into working knowledge instead of an open question — or graduates into a Blog post once there's a synthesis worth writing down.