In 2 months, we went from 3 weeks to 3 days per screen migrating the Carrefour app (3 million users, 1 billion in revenue) from native Android to React Native with AI. To achieve both high-quality code generation and sufficient velocity, my team and I drew inspiration from other engineering domains, notably systemic defect analysis.

As the Tech Lead for this project, in this talk I will:

- Describe our overall quality process: our aggressive definition of a defect, how we map weak points, and how we measure the improvements they bring.
- Provide concrete examples of issues encountered during the Android to React Native transformation and how this process helped us fix them.
- Show how it reshaped the developer's role — shifting them from coders to true engineers.

You'll leave with: a defect-driven framework you can apply, a methodology for measuring AI code quality at scale and a new lens on how AI changes what "senior engineer" means in 2026.