← Back to Work
Iberia (via Alten)
AI QA Architect designing how AI-assisted quality engineering works inside an airline
Context
Quality engineering for AI systems is a different discipline from quality engineering for deterministic software: the same input does not produce the same output, so the unit of verification moves from assertions to distributions. The work is defining what that means in practice for an airline — which behaviours need to be guaranteed, which only need to be bounded, and how you prove either one to people who are accountable for the result.
SCOPE
- Designing AI-assisted test generation so coverage grows with the product instead of lagging behind it.
- Building evaluation harnesses for agentic and LLM-backed features, with regression suites that run on every change.
- Defining acceptance criteria for non-deterministic behaviour: thresholds, tolerances, and what constitutes a failing run.
- Instrumenting pipelines so quality signal is visible continuously rather than at release gates.
- Working across engineering and QA teams to move AI testing from ad-hoc experimentation to a repeatable practice.