You inherit a large UI automation suite that is frequently flaky and slow. How would you triage and stabilize the suite over the next 4–8 weeks? Describe the practical steps, metrics you would track, and trade-offs you might make.

When I inherit a flaky, slow UI suite I start by measuring a baseline in week one: flakiness rate, average test runtime, and CI pass rate (for example, 30% flakiness, 8h suite runtime, 70% CI pass). My immediate moves are triage and containment — tag and quarantine tests that cause >50% of failures, add minimal retries where failures are environmental, and run the suite in parallel to get faster feedback. Over weeks 2–6 I rewrite brittle UI asserts, increase API-level tests (moving ~40–60% of checks off the UI), and introduce stable locators and test data fixtures. I track flakiness %, mean time to green, and total runtime; my target is <5% flakiness and cutting runtime to under 2 hours in 4–8 weeks. Trade-offs: I accept temporarily reduced UI coverage and freeze new UI tests to focus on stability.