CourseAI / ML Engineer: The Complete Course · Module 9: Serving and Inference · part 59 of 93
Part 59 · Module 9: Serving and Inference

Incident: One shared filter stopped every model

15 min read·
Module 9 · Incident file · About 30 minutes · Needs: Lesson 2 (admission control, timeouts, the fallback ladder, prepare)

Everything below about the incident comes from Google Cloud's own incident report, quoted where it matters. The report does not describe the safety filtering service's design, so this file does not either.

The rest of this course is yours to keep

This course is bought on its own, once, and stays readable afterwards, including the parts added to it later.

Or open everything with a plan