Lab project

PodQApath

An agentic release auditor for QA teams. It connects Jira, GitHub, and AI to reconcile what tickets claim against what code actually shipped, and surfaces real risk before release.

Demo: auditing a release end to end.Elixir/Phoenix backend, React frontend

The Nurse System

PodQApath is built on a QA philosophy from six years in IoT property automation: meet Karen. She's an ICU nurse coming off a 13-hour shift, phone at 13%, standing at her front door. She doesn't care what changed in your sprint. She cares that her door opens. A failed test isn't a red badge on a dashboard; it's a person locked out of their home.

So releases get a three-tier risk model: Tier 0 is cosmetic, Tier 1 is non-safety device behavior, and Tier 2 (door locks, HVAC, access control) always requires human sign-off. The AI does the tedious reconciliation; humans keep the judgment calls that matter.

Why it's here

This tool runs on the same principle I bring to marketing: connect to what the system actually does, not what the ticket says. In marketing terms, measure what actually produces leads, not what looks good in a report.