Engineering notes, written slowly
I build applications, write about AI, APIs and system design - the silent bugs, the wiring, and the boring parts that quietly hold everything up.
Keeping an AI guardrail release honest
For an AI feature with guardrails, releasing it carefully requires two disciplines: coordinating multi-service promotions with evidence at each step, and keeping the validation environment honest enough that human feedback can be trusted.
What AI features force you to specify
When you add an AI layer to a retrieval feature, data decisions that developers used to make by instinct become explicit contract requirements. Live demos are often the first place you find out which ones are missing.
The guardrail is the feature
An agent that finds every reference to a feature flag and plans its removal is the easy half. The rule about which flags it is never allowed to propose removing is what makes it safe to leave running, and what let the job move to the person who actually wants it done.
Bad acceptance criteria ship as bugs
Ambiguity in acceptance criteria doesn't disappear when implementation starts. It becomes a decision made without full context. The lineage problem, staleness semantics, product positioning: caught before a sprint begins, they're conversations. Caught after, they're migrations.
The demo is the test
A feature that passes every test can still be broken in ways the test suite is structurally unable to see. Four kinds of blindness that only surfaced once real stakeholders started using an AI booking assistant in a room.
Codebase health is a lagging indicator
A tooling migration, some dependency bumps and a telemetry schema. None of it reaches a roadmap, and all of it decides whether a codebase is pleasant to work in three years from now.