Essays on AI, engineering, and the assumptions worth examining.
Agentic AI / Recursive self-improvement / Research
An analysis of recursive self-learning in AI agents through Dream-RSI: how replay improves exploration, what the results show, and where self-improvement falls short.
Read article ↗: Recursive Self-Learning in AI Agents: What Dream-RSI Changes9 min readAI-assisted development / Documentation / System ownership
AI can write and review a change while human understanding falls behind. A documentation agent could help preserve the knowledge teams need to operate their systems.
Read article ↗: When AI Writes the Code, Who Understands It?7 min readWhen AI moves from making suggestions to taking action, the question becomes how much responsibility to give it—and where human judgment belongs.
Read article ↗: Giving AI the Authority to Act7 min readGood engineering is often the reduction of unnecessary possibility.
Read article ↗: Constraint1 min readThe best infrastructure disappears into reliability. Most users never notice it exists — only when it fails.
Read article ↗: Invisible Systems1 min readA mature system is not the one with the most features. It is the one that behaves predictably under pressure.
Read article ↗: Operational Calm1 min readWhy infrastructure stops matching intent, how drift enters production systems, and how validation layers reduce surprise.
Read article ↗: Terraform Drift and the Illusion of Stability1 min read