WRITING
Practical notes on building software, measuring its behavior and putting AI to work.
AI Engineeringarticle
A practical boundary for AI-assisted engineering: code evaluates the evidence, while AI explains changes and proposes what to investigate.
Engineering Reliable AI Systems, part 1
AI Engineeringarticle
A practical look at the point where an AI feature stops being a demo and starts needing repeatable evaluation.
Engineering Reliable AI Systems, part 2
AI Engineeringarticle
Tool calling can solve a surprising amount of useful AI work without introducing an autonomous agent architecture.
Engineering Reliable AI Systems, part 3
AI EngineeringarticleUpdated Oct 28, 2026
Design a secure cross-team .NET agent integration with typed HTTP clients, authorization, deadlines, task lifecycle and traces, before introducing AI Mesh.
Building Enterprise AI Systems, part 4
AI EngineeringarticleUpdated Oct 28, 2026
A practical .NET design for read-only incident investigation with typed evidence, per-tool timeouts, partial failures, and tests, before adding a mesh.
Building Enterprise AI Systems, part 3
AI EngineeringarticleUpdated Oct 28, 2026
A practical architecture for distributed AI incident investigation, covering agent boundaries, MCP, A2A, security, tracing, failure handling, and when not to use a mesh.
Building Enterprise AI Systems, part 2
AI EngineeringarticleUpdated Oct 28, 2026
A practical look at AI Mesh, the problems it addresses, and how it relates to MCP, A2A, and AI gateways without adding unnecessary architecture.
Building Enterprise AI Systems, part 1
AI Engineeringnote
A small experiment in making AI-assisted code review narrower, evidence-based and testable instead of asking a model to review everything.
AI EngineeringnoteUpdated Oct 7, 2026
Adding orchestration, selective context gathering and a structured output contract around a narrow .NET review skill.