A retry is not a fix
If a test is flaky, something is genuinely non-deterministic — a race condition, a shared fixture, an assertion against a promise the app never made. Adding a retry hides it and teaches your team to ignore red builds.
About
I'm a QA engineer and SDET, and I've been doing this for fourteen years. I started in manual testing on desktop and IVR systems, moved into data and ETL validation where a wrong number means someone's payment is wrong, then into automation — web, then mobile, then desktop, then decentralised applications where almost nothing behaves the way the documentation promises.
Along the way I've led small distributed test teams, built frameworks from nothing, and spent a lot of time on the specific problem of making a test suite that a team still trusts a year later. That last part is the hard bit, and it's mostly what people hire me for.
These days I also build products. I take on feature work and whole applications, usually with an AI assistant in the loop from first prototype through to release. I like the combination: knowing how to break things well makes me better at building them, and shipping my own software keeps me honest about whatwell-tested actually has to mean when there's a deadline.
I'm based in Mérida, Mexico, which puts me in a full working-day overlap with teams across the US and Latin America. I've worked remotely with distributed teams for most of the last decade.
At a glance
How I think about it
If a test is flaky, something is genuinely non-deterministic — a race condition, a shared fixture, an assertion against a promise the app never made. Adding a retry hides it and teaches your team to ignore red builds.
I'd rather have fifteen tests that cover the journeys your revenue depends on than four hundred that assert a button is still blue. Risk decides what gets automated, not a target number.
I write test code to the standard of production code and review it the same way, because a suite that becomes unmaintainable gets deleted — usually about a quarter after the consultant leaves.
The cheapest bug is the one caught while the feature is still being designed. I'd rather be in the requirements conversation than handed a release candidate on a Friday.
Experience
Employers are described by sector rather than named — current clients and roles stay confidential. Happy to talk specifics on a call.
2025 — Present
Early-stage startup · confidential
Own the quality layer end to end — test automation, CI and release verification — and build product features alongside the team.
2024 — 2025
B2B wholesale commerce platform · enterprise retail SaaS
Release readiness across multiple environments; automation for web and mobile.
2022 — 2023
Decentralised messaging platform · Web3
Test architecture and automation for web, desktop and mobile clients.
2021 — 2022
Global software consultancy
Cypress framework strategy, API automation, SQL and NoSQL validation.
2018 — 2020
Digital transformation consultancy
Led a distributed test team while owning ETL validation tooling.
2015 — 2018
Insurance data programme · via consultancy
ETL interface testing and SQL-based data validation at scale.
2014 — 2015
Beverage distribution startup
Built the QA process from nothing; led testers for a major beverage brand.
2012 — 2014
Nearshore software services
Desktop and IVR application testing; SQL query development.
Education
2013 — 2014
Universidad Tecmilenio
2005 — 2009
Universidad Autónoma de Yucatán
Toolkit
Get in touch
The best place to start is a conversation about what you're building and what's getting in the way.
Email — scheduling link coming soon