Topic – From Flaky to Bulletproof: Teaching AI to Smell Bad Test Patterns

Flaky tests are the silent killers of engineering productivity. Everything passes locally, but CI tells a different story. Pipelines fail randomly, teams hit “re-run” out of habit, and over time, trust in the test suite erodes.

But here’s the truth — flaky tests aren’t random.

They follow patterns.

In this session, we’ll break down the most common causes of test flakiness — from missing awaits and timing issues to shared state and environment dependencies. These aren’t one-off mistakes; they’re repeatable anti-patterns that show up across codebases.

Then we take it a step further.

What if we could teach AI to recognize these patterns automatically?

Instead of relying on manual code reviews or post-failure debugging, we can use AI to “smell” bad test code before it reaches CI — flagging risky patterns, suggesting fixes, and improving test reliability at scale.

By the end of this talk, you won’t just understand flaky tests — you’ll know how to eliminate them systematically and build test suites your team can actually trust with AI