QA Specialist
I break things on purpose so they don't break on their own.
The same suite that runs on every push to this repo.
I spent five years auditing operational processes, costs and ERP data. That job left me with a genuinely useful obsession: finding where something breaks before it reaches anyone.
I do the same thing with software now, as a QA Specialist at Light-it. I design test plans, validate APIs and document defects with full traceability. The automation side I've been building on my own, in personal projects. I get involved early in development, not once it's too late to fix anything.
I use AI every day to speed up the tedious part of the job. The judgment is still mine.
Sep 2026 – Present
Aug 2024 – Aug 2026
Freelance
Sep 2021 – May 2026
2024 – 2025
Functional, API and automated testing · Test case design · Postman and Swagger · Agile methodologies · Traceability and continuous improvement.
2018 – 2019
Object-oriented programming · Algorithms · Desktop application development.
Courses marked “in progress” are the ones I'm taking right now. The label changes when the certificate lands.
I don't use one tool for everything. I pick per task, which is exactly the criterion I apply to choosing any testing tool.
Breaking down user stories, spotting ambiguity in acceptance criteria and building edge-case matrices. I tried Gemini for the same work and stayed with ChatGPT: it holds the context of a long requirement better and returns less redundant cases.
Inside VS Code, for writing and refactoring automation code. Autocomplete and scaffolding, not design decisions.
I built this site with Claude Code: architecture, components, the E2E suite and the CI pipeline. The result is the evidence — including the tests that verify the page doesn't break.
A concrete example: this site's content exists exactly once, and it's typed. Spanish and English validate against the same contract, so if one language is missing something, the build fails. Both PDF CVs are generated from that same source. It's quality control applied to my own information.
Two user stories, 22 test cases generated by an AI assistant, and a written verdict on every one — kept, fixed or rejected, with the reason. 15 made it into the Playwright suite. Four are marked as expected failures, because running them found real defects in the application under test: a negative amount runs the transfer backwards, and the amount field accepts cents that the app then silently drops. The point is not that AI wrote the tests. It is that the review is written down and anyone can check it.
End-to-end automation against the-internet.herokuapp.com: successful and failed login, dynamic element handling and chained workflows. Page Objects on a shared BasePage, Playwright fixtures for dependency injection, and data-driven cases that cover several failure scenarios from a single test. Runs on GitHub Actions on every push and PR, with the HTML report as an artifact.
API tests against JSONPlaceholder using Playwright's request fixture: status code verification and response content validation on GET and POST. Work in progress — growing towards PUT and DELETE, resource-based organisation with an API client pattern, and combined UI + API testing.
A tool that solves a problem of my own: sending my CV to the openings that show up on LinkedIn, and knowing a month later which opening each reply belongs to. A Manifest V3 extension spots the addresses in the feed —including the ones LinkedIn turns into mailto links— and opens a panel with company, role and language prefilled from the post; a local Python server builds the email from bilingual templates, sends it over SMTP and labels the Gmail copy over IMAP. It stores the whole post, so a search finds it later by any word it contained, and a matcher pairs replies using Gmail's conversation id. Dependency-free backend, 79 tests with their own fixtures, and CI that also fails if a credential ever gets committed.
Static bilingual site with typed content as the single source of truth, a Playwright E2E suite (language parity, accessibility, responsive) and CI running typecheck, build and tests on every push. Both PDF CVs come out of the same content.
I work as a QA Specialist at Light-it. This site stays up as a reference — reach out if you want to connect for any other reason.
References available on request.