Williams Liu.
A QA who learned to ship.
10+ years of software testing across gaming, education, media, blockchain, and AI platforms. Bringing QA engineering rigor into AI agent workflows — built a battle-tested dual-layer QA Orchestrator wiring ticketing systems, test-case management, comms tools, and test reports end-to-end. Specialties: requirements analysis and decomposition, test scenario design, and user-perspective testing.
EXPLORE →test once,test everywhere.
Career archive
Recent to early — five chapters of practice, refined over time…
-
QA Engineer (AI Agent Architecture)
Create Intelligens Inc. — SET GroupSolo QA ownership across six product lines — member-engagement platform, LINE CRM, company site (Next.js monorepo), AI virtual-human system (Backend / Brain AI / TTS), and early-stage PoC validation.
- Self-designed dual-layer QA Orchestrator: each layer holds its own test memory and workflow playbook. Ironclad governance — release-gate confirmation, self-learning from defects, sensitive-data exclusion from version control, Prod-environment guardrails — auto-synced across agent configs. Git-versioned, auditable, reproducible.
- Stood up and operates the in-house MeterSphere v3 platform via AI cowork (Test Plan / Case / result writeback / step writeback / screenshot upload / batch update).
- Wired the full test loop with AI cowork: Jira ticket analysis → test plan & cases → mixed manual + automated execution (Playwright) → writeback to test cases and Jira → final test report. Plus QA Filter, QA Weekly Bot (Slack incoming webhook), progress autosave, and work log.
-
Software QA Engineer
Bityacht LimitedBuilt from 0 → 1 with experience plus AI cowork: manual test cases and plans, API test framework (Python + pytest with Cursor), Playwright web automation prototype, full test docs and reports.
- Reached ~34.9% API test coverage in two months without prior API docs.
- Layered OWASP cases into the API suite per security requirements — authorization flaws, SQL injection, XSS.
- Qase as TCM, pytest + Cursor as the API automation backbone.
-
Performance QA Engineer → QA Deputy Manager (managing up to 4)
YATAI MANYU TECHNOLOGY CO., LTD. (FameEX / Top.one)QA for Top.one — a centralized blockchain trading platform: user info, KYC, address book, wallet assets, platform banners and announcement CMS.
- 0→1 testing across web / iOS / Android (Product A → Product B extension).
- Pre-spec meetings: risk + capacity assessment, cross-team alignment, logic confirmation.
- Authored test plans and cases as foundation for automation peers (regression / smoke).
- Production triage, defect localization, intake into dev workflow.
- Promoted to Deputy Manager — managed a 4-person QA pod.
-
Software QA Engineer (EdTech, e-commerce, online learning)
OnePath Education Co., Ltd. (OneClass) + ITER TECHNOLOGY LTD.QA for OneTeacher (teacher-side) and OneClass (student-side) on web / iOS / Android — class schedules, booking, replay, post-class feedback, course-trial reservation. Spanned order-system rebuild (payments), back-office rebuild, product-customer points consumption.
- Shipped order-system rebuild; the new system delivered ≥30% revenue lift in its first month.
- Built a Postman-driven API monitor scheduling weekly health checks on critical endpoints.
- Full-stack QA (web PC / web H5 / Android / iOS) + Jenkins deploys + GitLab version verification.
-
Game QA · Test Lead · Game Designer
Gamania Digital Entertainment → Chinesegamer International → Next Animation Studio → HAPPYTUK → Digiforce → BugstechThe formative years of my QA discipline.
- Bugstech (2015-2017): Managed a 13+ person card-game QA team (3-shift, peaking at 50 people) — staffing interviews, KPI design, annual reviews, shift scheduling, cross-team coordination, on-site vendor exchange in the Philippines.
- Digiforce (2013-2014, 1y 9m): Web editor — daily curated iOS limited-time-free and chart-ranking titles, wrote quick reviews and published to the company site and the in-house "Limited-Time Free Games" app. Also planned game-strategy companion apps end-to-end, coordinating with design and engineering on UX, dev, testing, copy, and maintenance.
- Gamania Digital Entertainment (2009-2011): Lead QA on MapleStory for 2 years; supported version tests for CSO / KartRider / Dragon Nest / Bnb / DNF; acted as test lead for 3 months during a re-org.
- Earlier rotated through game planning, game analysis, set-top box game dev — broad surface area, all feeding back into test instinct.
Gamania
Chinesegamer
Next Anim.
HappyTuk
Digiforce
Representative products from past roles
Products I led, tested, or planned during my career.
- 2026/3 — Present
QA Agent Operating Framework
Platform-layer governance for multi-product, multi-agent QA. Hub-and-Spoke topology (Orchestrator + N project agents), three-tier knowledge architecture, 8 iron laws + 26 self-learning rules + 12 heuristics, cross-layer sync tooling, three-tier secret-repo separation so the framework layer stays shareable while company config and credentials never leave their own boundaries.
6 LAYERS QA Agent Operating Framework- A Agent Topology Orchestrator + Project Agents
- B Knowledge Architecture Global / Project / Agent memory
- C Governance Iron · Self-Learning · Heuristic
- D Sync & GitOps push helper · pre-commit lint
- E Secret & Cloud L1 public / L2 private / L3 local
- F Tool Capability Issue Tracker · TCM · Chat · Secret
- 2023/3 — 2025/6
Top.one — Centralized Crypto Exchange
YATAI MANYU TECHNOLOGY CO., LTD. (FameEX)QA → Deputy Manager (led 0-1 launch, KYC, wallet, CMS testing)
- 2022/6 — 2023/3
OneTeacher / OneClass — Online Learning Platform
OnePath Education Co., Ltd. (OneClass)Full-stack QA across web / iOS / Android (teacher-side + student-side + order/payment system)
- 2025/8 — 2026/1
Bityacht — Crypto Exchange API QA
Bityacht Limited0-1 API test framework, 34.9% coverage in 2 months, OWASP cases
- 2009 — 2011
MapleStory — Online Game QA
Gamania Digital Entertainment Co., Ltd.Lead QA for 2 years; cross-supported CSO / KartRider / Dragon Nest / Bnb / DNF
Experimental projects
Between burning inspiration and burning time, I’d rather burn tokens…
- /01
DrawPicker — Lightweight Raffle Tool
● liveA browser-side raffle / random-pick utility — lightweight, no backend, supports custom candidate lists.
- /02
FakeSmith — Test Data Generator
● liveA Chrome extension born from a QA workflow — quickly generate, store, and export test data without leaving the browser.
- /03
WIL CRM — Takeout / Delivery Ordering System
◐ wipBuilt for a relative’s takeout / delivery afternoon-tea shop — an attempt at a multi-tenant F&B CRM customer-side storefront with merchant backoffice and order flow (Next.js 16 + Tailwind v4 + Postgres).
- /04
Spin Rhythm XD — Traditional Chinese Localization
◐ wipOngoing volunteer localization of the rhythm game Spin Rhythm XD into Traditional Chinese (Taiwan). More cultural adaptation than literal translation — driven through the community and online channels. A fan project for a game I genuinely enjoy.
Interests
Off-the-clock — for play, broader perspectives, and unexpected cross-pollination.
Skills
Theory in practice — testing methods, AI cowork, TCM, infra, comms.
Testing / QA
- Manual testing (black-box, boundary, negative, weak-network, regression, smoke)
- API testing (Python + pytest)
- E2E automation (Playwright)
- OWASP security cases (authz / SQLi / XSS)
- ISTQB methodology
- Test plan & case authoring
- Defect localization & tracking
AI Cowork
- Claude Code · Cursor · Copilot
- QA Agent platform framework (6-layer governance / Orchestrator + Project Agents)
- Agent governance: iron laws / self-learning / cross-layer sync
- AI-wired test loop: Jira → cases → exec → writeback → report
- Prompt design & eval mindset
TCM / Bug Tracking
- Jira · MeterSphere · Testrail · Qase
- Mantis · Bugzilla · Notion · Lark
Tools / Stack
- Postman · Hoppscotch
- Stream · Fiddler · Charles (mobile traffic)
- Chrome DevTools (console / network / ws)
- GitHub · GitLab · Jenkins · Docker · Linux
- Zeabur · Cloudflare · GCP
- Figma · Miro · SwaggerHub · Axshare
Collab / Comms
- Slack · Lark · HackMD · Notion · Google Drive
- Cross-team coordination / 50-person rotating-shift team leadership
Languages
- Chinese / 中文 Native
- English / 英文 Conversational
- Taiwanese / 台語 Intermediate
Notes.
Occasional notes — when something is worth jotting down.
Let’s talk.
Open to conversations on QA, AI workflow, design systems, and how to ship the boring fix. Email is fastest.