Took a business-critical legacy app from effectively zero coverage to ~6,000 automated tests.

Problem

A large, business-critical PowerBuilder application had no maintainable automated test coverage. Every release leaned on manual QA, feedback was slow, and any change to the legacy code carried real risk of silent regressions.

Approach

I helped design and build a modern end-to-end testing platform in .NET and C# that could drive a legacy UI never built for testability. Rather than chase total coverage, we prioritized the flows that mattered, identifying about 200 high-value regression tests as a dependable safety net, then expanded to roughly 6,000 automated cases spanning UI-driven and data-validation scenarios.

Outcome

The system went from effectively zero reliable coverage to a structured, repeatable testing model. Manual QA cycles dropped, and developers gained the confidence to modify legacy code without flying blind.

Reflection

Pragmatism beat purity. A small set of high-value tests bought most of the confidence, and made the case for funding the larger suite.