Engineering Blog
Practical insights on Playwright, self-healing tests, AI-driven QA, and scaling engineering teams. Real lessons from 20+ years in quality engineering.
162 posts
An agent can execute different arguments from the ones you approved. Pause the workflow, mutate one field, and test the actual write against the reviewed snapshot.
Read articleAn AI judge should pick the same answer after its position changes. Build a swap test, verify winner mapping, and separate order sensitivity from ordinary variation.
Read article
Keeping every error trace helps debugging but distorts failure percentages. Test sampling policies with a synthetic workload and verify the denominator behind each dashboard metric.
Read article
Two hundred AI evaluations may cover only twenty questions. Separate repeatability from coverage, pair model comparisons, and preserve case groups in uncertainty estimates.
Read articleDeleting a document is only the start. Test retrieval, cached answers, citations, and conversation history before declaring an AI assistant has stopped using it.
Read articleA running AI server may not be ready for requests. Test startup barriers, traffic routing, and recovery with a negative control that proves your readiness check works.
Read articleA practical AI bug-report checklist: preserve the request sequence, fixture revisions, system state, and failed assertion before a passing retry erases the trail.
Read articlePage 1 of 23
Get weekly insights on AI-driven QA, engineering leadership, and automation strategies.
No spam, ever. Unsubscribe anytime.
Built with Vite, React, and Tailwind CSS.
View Source