AI Operations

The fix for a poison-pill row re-billing a model every sixty seconds forever was not a rewrite. It was one INT column, capped at five, incremented atomically in the database. The fix itself shipped with a gap, and that gap is the actual lesson.

Eight rungs mapping what shipping production Agentic AI on AWS actually takes, marked honestly as covered, strong, flagship or gap rather than filled in to look finished.

A post framed fine-tuning as a 2026 interview trap. Here is the AWS version: the ladder before you touch weights, the four job types and three serving paths Bedrock forks 'fine-tune' into, and what each one costs, priced today.

Part 3 left me a label-free signal that responds to a real regression. It does not come with a threshold. Here is where the line actually goes, what a false page costs, and why the reflex answer computes to a negative number.

I could not detect a deleted guardrail with an LLM judge, a judge-free assertion, or six label-free signals. Three instruments, one null. The instrument was never the problem: shorten the source and the same gate goes from p = 1.000 to p = 0.0020.

I deliberately broke my summariser's prompt, then failed to detect it two ways: with an LLM judge over a golden dataset, and with a judge-free deterministic assertion. Removing the judge changed nothing. Here is the calculation that would have told me first.

I built an LLM-as-judge eval on my own blog and got a suspiciously perfect 16/16. Here's the three-round test I ran before trusting that number: single-variable corruption, and a self-consistency check the research says most teams skip.

AgentCore long-term memory has a read-after-write gotcha the docs skip: a direct BatchCreateMemoryRecords write returns 201 and stays unsearchable for 15 to 30 seconds. Measured, with the two-tier model that explains it.

The most expensive AI agent failures don't throw an error, they hide. One ran at a six-figure-a-year rate for days while every dashboard stayed green, because the signals that catch it are the ones nobody watches. The two instruments you are missing.

Strands' BedrockModel ships with prompt caching off. Two kwargs turn it on, one per-model gotcha catches you, and a 10-turn driver measures 99.9% and 99.8% hit ratios against an 8,156-token production system prefix. The usage block proves it in seconds.

Three findings from a real customer engagement on AWS Bedrock: what a load test was actually doing, why p95 latency was 45 seconds, and the prompt-caching default that costs every team money. Plus the three CloudWatch metrics that catch all three.

A personal Bedrock-adjacent build that went through three iterations and an architecture pivot. Five lessons that map directly to production AWS AI work.

Real cost breakdown of running an AgentCore agent: prompt caching savings, when to use Nova Pro vs Claude Sonnet, PriceClass_100, idle timeouts, and how to set alarms before your bill surprises you.