QA Engineer for PWA + API

Description: We've built an AI assistant platform for real estate agents. It autonomously handles email triage, lead tracking, task management, scheduling, and deal-based conversations — with a mobile-first web app (activity feed, multi-channel chat with voice input and output, image attachments, approvals, and lead management). It currently runs a real brokerage's daily operations. Before we onboard our next clients, we want a technically skilled tester to try to break it. How this works: we onboard you exactly the way we onboard a real client — you experience our actual setup process, then get a fully sandboxed test account loaded with realistic synthetic data (no real customer data anywhere near you). You'll receive an architecture briefing and an API map so you understand how the system is put together, and then you test it from the outside, both through the UI and directly against the API. What you'll do: Structured test passes across every surface: login and roles, feed, chat channels (create/rename/archive), voice input and read-aloud, image and file uploads, approvals, and lead management — on desktop and mobile browsers, light and dark themes Role and access-control testing with three account types (owner / admin / member): verify scoping through the UI and by probing the API directly — confirm that what should 403, actually 403s Exploratory testing: edge cases, error states, network drops, oversized uploads, rapid double-actions, stale sessions A comprehension pass: tell us what you think happened on each screen — we're testing whether the product explains itself to non-technical users (our customers are realtors, not engineers) Clear, reproducible bug reports: steps, expected vs. actual, screenshots, and console/network evidence where relevant What we're looking for: A QA or software development background — you can diagnose from the outside: browser console, network tab, API request/response payloads, and describe failures precisely ("the PATCH returned 200 but the UI re-rendered stale state" — that level) Comfort testing against a documented API surface Experience with web app / SaaS QA; PWA and mobile-web familiarity a plus Strong written English for reports NDA required; all work is against the sandboxed test tenant only Bonus: any real-estate industry familiarity Engagement: roughly 15–25 hours over 1–2 weeks. A structured test plan is provided, plus dedicated exploratory time. Hourly ($25–45/hr depending on profile) or fixed-price — propose what suits you. Strong performers get first call on a follow-up round of native-app (TestFlight) QA later this quarter. In your proposal: one or two lines — what's the best bug you've ever caught that the developers had missed, and how did you find it? That's paste-ready. Post it, pick from bios + that answer, and while proposals roll in, the machine's pre-day-one queue is ready when you are: stand testco up persistently + generate the architecture one-pager — say the word and both prompts ship in one message. describe what im looking for in two sentences Looking for a QA engineer with a developer's eye — someone who can test a PWA + API platform end to end, probe endpoints directly, read console and network errors, and write bug reports precise enough that a developer knows exactly where to look. You'll be onboarded like a real client into a sandboxed environment and paid to break an AI assistant platform before our next customers ever touch it. what do you think about all of this Good bones — the summary reads exactly right. Four fixes before you hit Finalize, one of which really matters: 1. Budget: $5–15/hr will get you the wrong person — this is the one that matters. The entire spec is "developer's eye, probes APIs directly, reads payloads precisely." That skill set doesn't exist at $5–15/hr; that band buys click-through-screens manual testing with templated bug reports. You'll drown in 50 low-quality proposals and the person you actually described won't apply. Set $25–45/hr (what we scoped when the bar went up). The engagement is ~15–25 hours — the total difference between $15/hr and $35/hr is maybe $400, and it's the difference between finding the leads-gating class of bug and finding "button color wrong." Cheapest $400 in the project. 2. Skills tags are misaimed. "Selenium / Selenium WebDriver / Automated Testing / Microsoft Azure" — Azure is irrelevant, and Selenium tags attract automation-framework engineers, not the manual+API explorer you want (you're not asking them to write automation). Swap to: Manual Testing, API Testing, Web Application Testing, Bug Reports, Postman (and "Exploratory Testing" if offered). Tags drive Upwork's matching, so this changes who even sees the post. 3. Scope: "1 to 3 months" oversignals. The engagement is 1–2 weeks / 15–25 hours. Set it to the smallest duration option — a months-long "Medium" scope attracts people fishing for long contracts and makes short-engagement specialists skip it. Intermediate level is right (arguably Expert given the bar, but Intermediate + your rate band is fine). 4. Screening questions — add the one, since that section is sitting empty: "In one or two lines: what's the best bug you've caught that the developers had missed, and how did you find it?" That's your bio-confirmation filter, agreed and cheap. Also: make sure the full description (the long version I wrote) is in the description field, not just the summary — the summary sells, the description sets expectations (NDA, sandbox-only, comprehension pass, the API map). If Upwork only showed you the summary box so far, the next screen usually has the body. Fix those four, then Finalize. Worldwide is fine (strong technical QA is global; your NDA + sandbox design already assumes zero-trust). Post it and the clock starts — and the machine's pre-day-one pair (stand testco up + architecture one-pager) is one word away.