QA engineer: AI-assisted test automation and Playwright, mortgage advice platform (part-time UK/EU)

We build the platform a group of UK mortgage and protection advice firms runs on. Multi-step client forms with conditional sections, a case pipeline, credit search and sourcing results, PDF output, roles and permissions. It is in production and used daily by advisers. There is no QA on the team yet. The work Test features on staging before they ship and write bugs a developer can reproduce on first read Build the Playwright suite from the handful of tests we have, covering the flows that matter most Use AI tools (Claude, Copilot, Cursor or similar) to generate and maintain tests, test data and scripts, and show us what works and what doesn't Automate the boring parts: test data setup, environment checks, regression runs in CI Decide with us what belongs in end-to-end tests and what should be a unit or API test instead Check the API directly when the UI hides the real problem You 3+ years of QA on web applications with complex forms, roles and data Playwright on a real project, TypeScript preferred Already using AI coding tools in your day-to-day testing work, with a view on where they help and where they get it wrong Comfortable writing scripts (TypeScript, Python or bash) to automate your own workflow REST API testing Comfortable being the only QA in a small team UK or EU based. NDA before access. Bonus: pytest, Vitest and React Testing Library, SQL, fintech or lending experience. Terms Hourly, 15–20 hours a week. Paid three-week trial: exploratory testing of one feature area plus Playwright tests for its main flow, then a long-term contract if it works for both sides. Four hours overlap with UK time. Small team, decisions the same day, one Slack channel. To apply Start your proposal with "Playwright" and answer the questions below. Short answers are fine. Proposals without answers will not be read. 1. We want every field of a 40-field client form covered by Playwright end-to-end tests. Estimate the effort. 2.Our Playwright tests log in through the UI with two-factor authentication on every test. Pick the approach you would use instead and say what it breaks. 3.A colleague proposes running the Playwright suite against production with a dedicated test account, so we test the real thing. Agree? 4.The flakiest test you ever owned: what flaked, how did you prove the cause rather than guess it, and what fixed it. 5.Show us one thing you have automated in your own testing work using AI tools. What did the tool get wrong, and how did you catch it?