Hieu D.

I cut a UK fintech's regression cycle by 85% and scaled a SaaS test suite from 0 to 3,000+ automated cases at 98% daily pass rate. Senior SDET / QA Lead with 7+ years specializing in modern web automation (Playwright + TypeScript), data pipeline validation on Microsoft Fabric, and AI-augmented testing workflows. I work directly with founders, dev leads, and engineering teams to build automation right the first time — no outsourcing, no team handoffs. ISTQB Foundation Level Certified | GitHub: DTHieu95 | Live Allure dashboard with trend history (in portfolio). 🏆 Recent Wins - UK Fintech (data platform on Microsoft Fabric) — Built end-to-end data validation framework using Python + Pytest + Pandas, covering 4 data pipelines into 9 gold tables. Achieved 100% data accuracy validation between source and gold layers, catching schema drift before stakeholder dashboards. - UK SaaS (digital workflow platform) — Designed Playwright + TypeScript framework from scratch. Scaled to 3,000+ test cases at 98% daily pass rate. Cut full-suite execution from 14-15h → 10h via parallel execution and role-based authentication. Manual regression cycle: 1 week → 1-2 days. - UK Enterprise (legacy .NET 4 web app) — Replaced manual-only QA with Playwright automation. Reduced single-ticket regression from 1-2h manual → 15-20min automated (~85% reduction), including PDF invoice extraction and calculation validation. - APAC Tier-1 Telecom — Led QA team of 4. Selenium + Java + TestNG against a Siebel CRM-backed customer portal serving millions of subscribers. Defined test strategy and reported quality status directly to client stakeholders. - Open-source Playwright reference framework — fixture-injected POM, 6 projects (UI + API + accessibility with axe-core), 2-shard parallel CI on GitHub Actions, live Allure dashboard auto-published with trend history. Available in my Portfolio section. 🤖 What makes me different: AI-Native QA Workflow Most QA freelancers either fear AI or use it as a fancy autocomplete. I've built a structured AI workflow that goes past surface-level prompting: - Research-first authoring with Context7 MCP — pulls live library docs before any code change, keeping output aligned with current APIs (not stale training data). - MCP Playwright authoring — page snapshot + role-based locator discovery + self-verification in a controlled browser before commit. - Self-verification guardrail — every AI self-test must execute the actual committed code, never improvised alternatives — closing a subtle failure mode where AI 'passes' verification against drifted code. - AI-assisted CI failure triage — root-cause flaky tests from long Playwright trace logs in minutes, not hours. The result: faster delivery without the typical AI-generated tech debt. 🛠 Core Stack Test Automation: Playwright (TypeScript), Pytest, Page Object Model, Selenium WebDriver Data & Backend: Python, SQL, Pandas, Microsoft Fabric, REST API testing CI/CD: Azure Pipelines, Azure DevOps, GitHub Actions, Git AI Tooling: Claude Code, MCP integrations (Playwright + Context7) Languages: TypeScript, Python, Java, SQL 📩 Let's talk If you're building a SaaS, fintech, or data platform and need automation done right the first time — message me. I respond within a few hours during UK/EU business hours. Available 20-30 hrs/week, open to long-term engagements with serious teams.