Senior SDET  ·  AI and LLM Testing  ·  Test Architecture  ·  QA Automation  ·  Regulated Domains and Payments

Senior SDET, ten years owning test architecture, API and integration frameworks and CI/CD release gates for systems where reliability matters: regulated healthcare and pharma, airline retailing and payments, a payment service provider, fintech and market data. I work at the framework and infrastructure level rather than only writing tests, and I own what gets tested, at which level, and when a release ships.

Since 2023 that has included testing AI systems in production, two years of it on a regulated product: a RAG assistant where retrieval and generation are validated separately, prompts and responses are checked against approved sources, and evaluation runs are gated in CI. Golden datasets built against a rubric agreed with domain experts, versioned in Langfuse, with model outputs cached so regression is deterministic.

I review developer written tests, design harnesses that teams adopt, and push testing left into design and code review.

→ Full background, experience and competencies on mariusargatu.com/about

• AI and LLM testing: does the model answer faithfully, and can you prove it? RAG tested stage by stage rather than as a black box (knowledge base, retrieval, reranking, generated answer) · evaluation rubrics agreed with domain experts as the quality contract · LLM as judge calibrated on human labelled sets, judge agreement tracked as its own metric · IR metrics as diagnostics, not as the release signal · agents scored on tool selection and trajectory · prompt injection, jailbreak and red teaming
• Test architecture: the frameworks and runners under the tests. test strategy across teams and products, integration and system tests carrying most of the coverage (testing trophy, honeycomb) · services exercised through SDKs generated from their OpenAPI specs · consumer driven contract testing in CI · Testcontainers, service virtualisation and failure injection · model based testing from state machines (xState), property based and metamorphic testing
• Regulated domains and payments: the release gate is a decision, not a green tick. multitenant healthcare and pharma SaaS, IATA NDC order and offer flows, checkout and settlement, event driven payment paths tested for retries, idempotency, ordering and message loss

• Freelance, B2B contractor · Jan 2022 – present
  • Senior SDET, airline retailing and payments platform · Jul 2025 – Jul 2026 — quality strategy and release gates for the payments squad; Python and Pytest API/integration framework, Pact and Schemathesis in CI, Testcontainers, Microcks, WireMock and Toxiproxy for real and degraded dependencies; OWASP ZAP, SonarQube, Snyk and load thresholds as gates; an internal LLM debugging agent correlating failures with Datadog traces
  • Senior SDET, multitenant healthcare and pharma SaaS, regulated · Jan 2023 – Jul 2025 — evaluation system for a RAG assistant on Azure OpenAI, Copilot Studio and Qdrant; rubric, golden datasets and LLM as judge gated in CI; agentic paths on LangGraph; Kafka ingestion validated for schema, ordering and replay; MCP servers exposing reproducible test context; Playwright and TypeScript platform adopted across teams, ~1,000 permissions scenarios generated from one xState model
  • Test Architect, native Qt desktop product · Jan 2022 – Jan 2023 — Pytest API framework with a DAG based concurrent runner, 1 hour to ~5 minutes; patched Qt WebEngine to drive the embedded web surface over CDP, per test effort 4 days to 1
• LSEG (London Stock Exchange Group), Senior QA Automation Engineer · Mar 2021 – Jan 2022 — cross platform Screenplay framework in Python covering a market data SPA (Playwright) and Windows desktop client (WinAppDriver); streaming and delayed feeds validated against source data, not rendered values
• 2Checkout, QA Automation Engineer · Mar 2020 – Mar 2021 — Pytest framework for ecommerce, subscription billing and cross border payment flows; database diff tool for snapshotting and validating state at the data level; Elasticsearch indexing and relevance testing
• Amber, QA Tester → Senior QA Tester → QA Automation Engineer · Apr 2017 – Mar 2020 — end to end REST integration testing, k6 load validation, CI/CD on Docker and Kubernetes; built an in house Android and iOS device farm and the Appium and Python framework running on it
• Electronic Arts, QA Tester · Apr 2016 – Oct 2016 — end to end feature testing for AAA titles across platforms

• QARoom: a multitenant social platform built to demonstrate testability as an architectural property, using chaos engineering, contract testing, mutation testing, and model based and property based testing, all mapped to the boundary they defend.
• PayFlow: a payment system built by an AI agent from a frozen spec, zero hand written tests, guarded by a four layer verification pyramid (structural, behavioral, agent judgment, mutation). 73.1% mutation kill rate.
• Atlas: a testable broadband support agent (LangGraph + MCP), the runnable reference system for the "Evals Are Checks, Not Tests" series.
• TemporalMutate: semantic mutation testing over Temporal workflows, where an agent hunts for blind spots in a payment workflow's test suite and every verdict comes from replaying recorded histories in milliseconds, with no live server.

• AI and LLM: Azure OpenAI, Copilot Studio, LangGraph, Qdrant, RAGAS, DeepEval, Langfuse, promptfoo, Guardrails AI, MCP, Ollama, Claude Code
• Languages and APIs: Python, TypeScript, SQL, REST, GraphQL, gRPC, WebSockets, OpenAPI, AsyncAPI
• Test tooling: Pytest, Vitest, Playwright, Selenium, Appium, Pact, Schemathesis, Microcks, WireMock, Testcontainers, Toxiproxy, Hypothesis, xState, Pydantic, k6, Locust
• Security: OWASP ZAP, SonarQube, Snyk
• Test management: Azure DevOps, Jira, Xray, TestRail, Qase.io, Allure
• Platform: AWS, GCP, Azure, Docker, Kubernetes, Kafka, Pub/Sub, BigQuery, Azure Pipelines, GitHub Actions, GitLab CI, Jenkins, Datadog, OpenTelemetry, Grafana

“A test suite is a liability as much as an asset. Every test earns its place.”