Quality Assurance Architect

Agentic Quality Assurance · Test Automation Architect

With 7+ years across test automation, QA leadership and software delivery, I architect quality assurance for organisations: the SDLC, the strategy, the frameworks, and the teams.

My mission is to accelerate releases without compromising quality while cutting the cost of recurring engineering work: test management and test automation architecture across every platform. I design the architecture, build the frameworks and agents, and grow the culture that delivers it.

📌 Endorsements

CORE EXPERTISE

• QA Strategy & Leadership — Test Maturity Models, SDLC design, QA team formation, performance frameworks, supplier governance.

• Test Automation Architecture — Web, mobile (iOS / Android), desktop, API, database and email, unified under single cross-platform frameworks.
• Agentic Quality Assurance — Engineered an agentic Quality Assurance loop for autonomous test automation (journey mapping → authoring → coverage expansion → bug discovery → failure diagnosis), adversarial / red-team AI testing, and model-agnostic LLM orchestration.
• Open-Source Library Design — Published multiple Maven Central and npm libraries used in production testing ecosystems.
• CI/CD & DevOps for QA — Azure DevOps, GitHub Actions, GitLab CI, Jenkins, pipeline parallelisation, Docker, Allure / Calliope / Slack reporting integrations.
• Languages & Stacks — Java, TypeScript / Node, Python, Swift. Spring Boot, Playwright, Selenium, Appium, Cucumber, Retrofit & more.
• LLM Platform & GenAI Engineering — Fine-tuning, RAG, MCP servers, agentic CLI tooling, harnesses & skill development, and local / self-hosted LLM deployment (Open WebUI).

WORK EXPERIENCE

Quality Assurance Architect & Managing Consultant @spriteCloud (01.01.2024 - present)
spriteCloud is a consultancy providing software testing services & test results dashboards.

• Architect clients’ QA landscapes end-to-end: assess practices through Test Maturity Models (TMM), design Software Development Lifecycle (SDLC) models that embed an effective quality assurance discipline, and size the QA capacity each work stream needs.

Client Engagements

• citizenM — a long-term engagement where I owned quality assurance end-to-end and managed the 10+ engineer QA team I founded (full details below).
• Ahold Delhaize — designed an unconventional Windows-desktop-to-SQL end-to-end automation solution for a complex testing challenge Ahold’s QA team had long wanted to automate, partnering with them to make it work; the engagement secured two spriteCloud placements, and I supervised the consultants who scaled it.
• BlueHero (AI-powered applicant tracking system):
  • Designed and built a Playwright + TypeScript E2E suite of 186+ tests across 16 feature areas, reaching near-complete functional coverage of a bilingual (Dutch/English) staff portal — authored via my agentic pipeline in hours, not weeks.
  • Pioneered an AI-vs-AI testing methodology: an LLM simulates real employers through five stress-test personas to drive the product’s AI chat assistant, and an LLM-as-judge verifies the published job against every fact the simulated employer stated.
  • Authored an adversarial AI guardrails suite running categorised attacks against the assistant, with structured LLM verdicts covering pass/fail, severity, violations and recommendations.
  • Designed a two-stage GitHub Actions pipeline (core E2E in ~2 minutes, AI run in ~25 minutes) that auto-posts transcripts and verdicts to pull requests.
• Farmedvisie / Medicheck (two-portal medication-administration platform, healthcare):
  • Delivered a durable E2E suite covering 34 user journeys — 500+ tests across both portals, written against an intent-level API so it survives UI re-renders, completing a full run against the live acceptance environment in under 20 minutes.
  • Engineered a regression-sentinel layer of 200+ tests — each documents a confirmed bug and turns green automatically the moment its backend fix ships, with no manual re-test.
  • Helped surface critical security vulnerabilities across the platform.
• MR MARVIS: designed and rolled out a new SDLC aligning development, QA and release practices, delivered an easy-to-scale Playwright solution for developers, and built an AI Playwright Test Writer generating ready-to-run test scenarios from user journeys gathered in a spreadsheet.

Firm-level leadership

• Managing Consultant — stepped up to help spriteCloud restructure and grow: shaping roles, titles and compensation, designing the performance-review process, and building company culture through team events.
• Innovation Lead — led spriteCloud’s AI strategy, founding Quail, the firm’s internal LLM platform spanning fine-tuning, RAG, MCP servers, agentic CLI tooling, harnesses & skill development, and a shared hub model for LLM-assisted QA:
  • Led the design of AI-native test authoring workflows adopted internally by consultants.
  • Contributed to Open WebUI, bringing local, self-hosted LLMs into the firm.
  • Built an easy-to-deploy CV solution modernising consultancy CV practices (the system rendering this document).
• Thought leadership — spoke at BrowserStack × spriteCloud events: “Let’s (not) talk about AI” and “Limitations of LLMs & the CAPTCHA Problem.”
• Mentored 7+ consultants through technical and professional growth, and built performance-measurement frameworks and KPIs for my team.
  
  

Quality Assurance Solution Owner @citizenM hotels (01.03.2022 - 31.01.2026)
citizenM is a Netherlands-based hotel developer, investor, and hotel chain — a major spriteCloud client I served as embedded QA Solution Owner across a ~4-year engagement.

• Architect-scope ownership of quality across the organisation: audited the entirety of citizenM’s software development practices to produce a five-stage Test Maturity Model, then designed the centralised QA approach it called for and managed the transition.
• Founded and grew the QA team to 10+ engineers, raising quality standards across citizenM’s product lineup — hiring, training and managing them across several product teams, reallocating resources as workload shifted, and running a continuous performance-feedback cycle for career growth.
• Redesigned the software development lifecycle to enable a solid test strategy: analysed the existing model, designed a replacement compensating for its shortcomings, and guided citizenM & its suppliers through the migration — separating development, test and acceptance environments while integrating test automation into deployment pipelines.
• Led Quality Assurance through citizenM’s backend migration — owning strategy, budget and hiring, implementing a performance test suite (JMeter, OctoPerf), designing a data-migration audit, and defining the UAT strategy between citizenM & its suppliers.
• Architected unified, cross-layer test automation across citizenM’s backend services and website, verifying frontends, backends, databases, customer emails and the analytics layer against each other:
  • Established continuous integration on Azure DevOps, with a custom script enabling test parallelisation at the DevOps level.
  • Slack integrations: live pipeline monitoring, a results-posting bot with slash-command remote triggers, and Calliope.pro → Slack reporting.
  • Established a single framework that simultaneously tested frontends, middleware, emails, analytics and the database against each other for true E2E automation — scaled with the QA team I founded to cover 5000+ test steps.
• Sunset & migration (2025–2026): delivered the final UI automation framework and CI (Pickleib), built a universal OpenAPI-driven test generator with self-healing API tests for an ultra-low-cost solution, and automated the post-sunset account migration (email-client integration, bulk-migration reporting) — securing contract extensions.
  
  

Senior Test Automation Engineer @spriteCloud (01.12.2021 - 01.01.2024)
spriteCloud is a consultancy providing software testing services & test results dashboards.

• Delivered client engagements across GitLab CI, Knapsack parallelisation (Cypress, Jest, RSpec), automated vulnerability & security testing (ZAProxy, Burp Suite), Cypress & Jest automation, and automated bug reporting via Jira & Trello integrations.
• Published a Test Automation Template covering web UI and mobile UI test automation examples, built on my Pickleib library (Java, Selenium & Appium), capable of parallel test execution (multithreaded and/or via BrowserStack).
• Published an API Automation Template built on my API automation library Wasapi (Java, Retrofit2 & OkHttp3), capable of parallel test execution, request interception & advanced header management.
• Improved test result monitoring using Calliope.pro & Allure for more insightful reporting.
  
  

Software QA Automation Engineer @Getir (02.06.2021 - 01.12.2021)
Getir is a technology company that provides solutions for the delivery of goods in urban areas.

• Delivered web automation, backend automation & backend development with JPA and Spring Boot: testing infrastructure for developers & testers, Dockerised virtual courier services, easily triggered regression suites and backend services for other teams — including a candidate-assessment backend built from the ground up, used by Getir to evaluate engineering candidates.
• Single-handedly programmed the entire web UI regression suite for one of Getir’s sub-brands, with a scaling adaptation of the Abstract Factory pattern enabling the project to support multiple sub-brand domains.
• Set up CI jobs and used Portainer slaves to pick up available jobs from Jenkins.
• Prepared Quickstart libraries combining BDD (Cucumber & Gauge) and Page Object Model, allowing entire automation frameworks to be bootstrapped within minutes.
• Implemented parallel execution as standard on every project.
• Authored an experimental automation framework that automatically acquires all elements & attributes with auto-generated xPaths into a project JSON file — interpreted by the framework — allowing automation without any manual page inspection (the conceptual ancestor of my current JSON-repository-based frameworks).
• Helped over 50 QA engineers at Getir learn test automation through a three-week bootcamp, complete with examinations and Getir certification. Sessions were captured and are reused for ongoing onboarding.
  
  

Software Test Automation Engineer @Pixery (15.07.2020 - 02.06.2021)
Pixery develops next-generation mobile applications for content creators. Worked on both Funimate (a social video editing platform used by millions daily) & Impresso (an all-in-one video editing tool for influencers and businesses).

• Wrote functional and integration test cases gating each release.
• Provided a flexible automation framework (Java, TestNG, Cucumber, Selenium, Appium, Jenkins CI) simultaneously usable across several mobile & web projects and API tests, with BDD scenarios writable by non-technical team members.
• Helped build a video-editing-engine regression solution performing frame-by-frame, pixel-by-pixel RGB comparisons across engine versions to detect visual regressions, with overlays that highlight the regressing effect directly on the affected frame.
• Developed Swift-based mobile test tools to measure video-editing-engine accuracy, alongside platform UI automation across the mobile apps.
  
  

Software QA Engineer @JustSnap (04.10.2018 - 01.07.2020)
JustSnap provides campaign solutions based on its in-house receipt-scanning ML technology.

• Participated in development, QA and BA processes of campaign projects for P&G, Philips and Pepsi.
• Responsible for quality assurance activities across 17+ campaign projects.
• Worked on cross-platform mobile applications, exploring mobile automation with Selenium & Katalon Studio and running sanity and regression suites.
• Implemented flexible web UI test automation solutions that were easy to set up, keeping pace with fast delivery schedules.
  
  

PROJECTS & OPEN SOURCE

Programming is my passion — I build reusable modules and publish them as open source under the Civitas Cerebrum ecosystem (npm, @civitas-cerebrum/*): an AI-native test automation stack that separates selectors from test logic, unifies execution across web / mobile / desktop, and gives LLM assistants the structured primitives they need to author reliable tests. It is the engine behind my client QA deliveries (BlueHero, Farmedvisie/Medicheck, MR MARVIS). Selected projects:

• Achilles — Autonomous Quality Assurance, shipped as a single npm install. One plain-English sentence — “complete E2E test automation of example.com” — and the agent owns the entire lifecycle: scaffold, crawl, journey-map, compose coverage, run adversarial bug-hunts, and produce a stakeholder deck, with no babysitting. Its distinguishing work is the harness engineering that makes an LLM agent trustworthy on real delivery:
  • A hook-enforced state ledger (30+ hooks that self-register in the agent runtime on install) gates every phase, pass and cycle transition — the agent cannot skip, shortcut or renegotiate scope; it either completes each phase or surfaces a blocker for human triage.
  • Tamper-evident, hash-chained ledgers plus self-protection guards stop the agent from editing the harness or forging its own progress state.
  • Schema-validated subagent contracts (JSON Schema) — every dispatched agent must return structured, validated output before the pipeline advances, with adversarial workflow-reviewer gates and attestation before any phase closes.
  • An 8-step selector-development guardrail safely injects test attributes using AST-diff and visual-diff verification across React / Vue / Svelte, reverting on any structural change.
  • Public-package contamination scanning keeps internal IP out of the published build; the harness itself is covered by 60+ hook tests.
• sql-client — Lightweight multi-engine SQL client (PostgreSQL, MySQL/MariaDB, SQLite, SQL Server, Oracle) with one API, a fluent query builder, typed results and always-parametrised queries — the data-layer oracle behind the Steps API.
• test-coverage — Static-analysis tool (TypeScript AST) enforcing 100% API coverage in test suites, with CI-friendly reporters.
• email-client — Zero-dependency TypeScript SMTP/IMAP client purpose-built for E2E flows (signup → verification → click-through).
• Pickleib — Utility library for Selenium WebDriver automation: driver setup, a clean interaction API, POM support, email testing and database connections.
• Wasapi — Lightweight Java library simplifying HTTP service generation and API calls with Retrofit — centralised configuration, no boilerplate.
• Napoleon — Java-based autonomous web agent: Selenium control through Pickleib, vision-based UI understanding and LLM decision-making, with a dedicated Nexus reasoning core — perceive → classify → decide → validate → act, fully in Java.
• EMTP (Expert Model Training Pipeline) — Python pipeline producing training datasets for domain-expert AI models: modular acquisition → enrichment → Q&A-pair generation.
• Easy Secret Santa 🎅 — Java app automating Secret Santa gift exchanges: pairs participants and sends personalised emails via SMTP. Built for my team, now used by multiple organisations.

PRODUCT & FULL-STACK PROJECTS

• BookHive — A full-stack bookstore with a React frontend, a Spring Boot backend and a MongoDB database, built as an automation-friendly reference app for exercising the Civitas Cerebrum testing ecosystem. Published as ready-to-run Docker images on Docker Hub.
• Vue / Android / iOS Test Apps — A family of multi-platform reference applications (Vue 3 playground, native Android, native iOS) used as public testing sandboxes for validating cross-platform automation frameworks.
• Food Planner — Lightweight personal web app exploring productivity and lifestyle workflows.

EDUCATION

Istanbul Technical University — Geological Engineering