Test Engineer

Phinomenon is built by browser and cloud veterans: our founders are behind FydeOS (a Chromium-based OS trusted by millions) and QingCloud (one of China's earliest API-first IaaS platforms, founded in 2012). Fresh off a healthy, eight-figure Series A raise, we have the runway to dream big without panicking about next week's payroll. With folks working from San Francisco, London, Beijing, Shanghai and Shenzhen, we are proudly global and friendly to your time zone, with overlapping hours so we can all actually talk. We ship fast and ship hard, but burnout is so last decade: we bake in wellness stipends, mandatory recharge days and flexible schedules, because great code comes from healthy lives. No timecards, no surveillance software, no weekend team-building. We'd rather hand you the best tools money can buy (Claude Max, ChatGPT Business and Cursor Ultra come standard) and trust you with them. We don't fetishise years of experience, degrees or big-tech logos, and fresh graduates are more than welcome: show us what you have built and how you think. If you want to build the nextgen browser that actually, you know, gets you, while still enjoying your evenings and maintaining upright posture, welcome home.

As a Test Engineer at Phinomenon, you'll be the reason Phi Browser ships confident, not lucky. You'll build the safety net under a product that releases every week on a moving Chromium base: automating real browser workflows across macOS versions and hardware, and pioneering how one tests agentic AI features, where correctness is a distribution of acceptable outcomes rather than a single expected value. You'll steer tools, including AI agents that test AI agents, so machines handle repetitive verification and humans do the thinking.

• Build end-to-end automation for real browser workflows (tabs, spaces, extensions, sync, updates) across macOS versions and hardware profiles
• Own regression suites for weekly releases and Chromium rebases; catch breakage before Canary users do
• Build eval harnesses for AI and agent features: golden datasets, LLM-judged assertions, distribution-based checks for non-deterministic outputs
• Red-team the agentic surface (prompt injection, adversarial pages, guardrail stress-tests) before the internet does
• Automate device-fleet testing: batch install, run, collect, and report across many machines with one command
• Turn every escaped bug into a permanent executable check, wired into our issue tracker and release gates
• Define quality dashboards and release-readiness signals the whole team actually trusts

• Real test-automation/SDET chops on desktop or client software; browsers and Chromium-family apps count double. We don't count years, we look at what you've built.
• Strong scripting in Python or TypeScript; fluent with Playwright, WebDriver, or accessibility-API-driven automation
• Hands-on experience testing on macOS: signing, updaters, permissions, entitlements, and their sharp edges
• A working understanding of how to evaluate LLM-powered features: evals over exact asserts, tracing, flake-versus-model-drift diagnosis
• A builder's approach to QA: you create tools and pipelines, not just test cases
• CI fluency and zero tolerance for flaky tests; you treat flakes as bugs, not weather
• Bug reports engineers thank you for: repro, expected, actual, evidence

• Experience testing Chromium-based products, browser extensions, or agent frameworks
• Familiarity with agentic-QA tooling (AI agents that plan and execute test flows)
• Hardware-lab experience managing a physical device fleet
• Contributions to open-source testing or eval tools
• You've filed a browser bug upstream and lived to see it fixed