Workforce quality assurance pipeline

The primary quality assurance system for a large provider workforce. It runs against the whole population, not a sample. When something goes wrong, an AI investigator reads the provider’s full relationship history and writes up the evidence. Issues become flags, three flags trigger an automated strike, and a person reviews it before it stands.

I built it and still own it: the decision rules, the AI investigators and their prompts, and the exemptions, through three major versions. Data engineering partnered with me to move it into the warehouse.