Stefan Florescu
Profile
8+ years in QA: 4.5 years on FIFA Mobile at Electronic Arts, then full-spectrum manual QA at Iron Sheep (scripted, exploratory and A/B testing) that grew into AI-driven test automation and game QA for a US social & sweepstakes casino platform. I build Playwright automation, re-calculate game results from the rules to prove payouts are right, and use AI agents (Claude Code) to make testing faster, with the judgement about what counts as a bug kept human. AI-assisted QA is where I want to grow next.
Experience
QA Automation & Game QA Engineer · Iron Sheep (Quality Analyst)
Remote · Dec 2022 – present- Full-spectrum manual QA (scripted, exploratory, A/B) that grew into AI-driven test automation and game QA.
- Built and maintained an end-to-end Playwright suite: ~350 tests in 61 spec files, 301 catalogued test cases, covering the platform and 44 game specs (Mines, Plinko, Tower, Crash, sports games).
- Set up Smoke and Regression runs with TestRail reporting and a test-run dashboard; cut the Plinko suite runtime by 26% with no new flakiness.
- Tested new games against their spec documents (~90 checks per game) and wrote GO / NO-GO release reports; a NO-GO with evidence on a feature that broke the play limits gave the team a clear basis for the launch decision.
- Wrote independent math verifiers that re-calculate every round from the rules: 4,000+ real rounds across 6 games, 0 maths mismatches; statistical sanity checks (chi-square, z-score, Monte Carlo) on hit rates and RTP.
- Ran release checks for batches of up to 66 games (currencies, banned wording, mobile and desktop views, max exposure) and audited the advertised max win across 144 games.
- Tested on real Android devices over wireless ADB and the Chrome DevTools Protocol; compared ported games with the original provider versions.
- Built and shipped the QA Companion Chrome extension (Manifest V3) used by the QA team: round log, balance checks, banned-wording scan, checklist export.
- Defined the QA process as 10 reusable AI agent playbooks (Claude Code) for spec-based testing, release checks and test scaffolding.
- Built Claw Machine with ZOOT’s AI game builder (Forge) from prompts and reference images; it became the template for two further games, Autumn Claw and Midnight Claw (all playable in the public ZOOT archive).
- Manual QA across the full range: scripted and exploratory (unscripted) testing, A/B testing, functional and regression testing.
Quality Analyst · Electronic Arts
Bucharest · May 2018 – Nov 2022- Tested FIFA Mobile for 4.5 years across its live releases: new-feature, regression and sanity testing.
- Wrote and maintained test cases in TestRail; logged and tracked bugs in Jira, with documentation in Confluence.
- Worked with builds from Perforce, ran risk analysis for test scope, and used data analysis (Excel) and Postman for investigation.
Skills
Education
High school diploma, Fine and Decorative Arts, "Ionel Perlea" Art High School, Slobozia (2008–2012)
Management studies, University of Agronomic Sciences and Veterinary Medicine of Bucharest (2012–2016, not completed)
Languages
Romanian: native · English: advanced