Skip to content
Andres Carreon

About

Hi, I'm Andres.

I'm a quality engineer who works remotely, with teams anywhere. For nine years I've made sure software does what it says it does.

I started at Accenture, where I led a QA team of 15 on enterprise systems. Then I spent four years building the test automation for a self-custody crypto wallet from zero across iOS, Android, desktop, browser extension and web. Today I work on quality and AI automation at a fintech startup.

I taught myself to code, mostly from official documentation. That's still how I learn: read the source, build something small, break it, and write down what happened.

Andres Carreon smiling on a pier at sunset, with a bridge across the bay behind him

Now

Updated September 2026

Day job
Quality and AI automation at a fintech startup: CI runners, evals for a support agent, and production code.
Building
jevvium, in the open. Next up: an Appium plugin and a Maestro export.
Measuring
Whether a decision model’s confidence can be trusted, by benchmarking it against the runs where it was right.
Writing
Notes on testing with AI, and essays on attention and building things alone.

After hours

I build products on my own time: nine shipped so far, two with paying customers, and one that took first place globally at an AI Tinkerers hackathon. They keep me honest about the other side of QA, where you are the one shipping and every shortcut comes back as a bug report.

Most of my AI tooling started there. An agent on a Mac mini triages the bugs in my own products, writes the ticket with a root cause and opens a pull request for the easy ones.

See the projects

How I work

  1. 01

    AI explores, tests decide.

    Let a model find the path through an app, but pass or fail comes from real checks, and CI runs plain code.

  2. 02

    Fast feedback is a feature.

    A slow suite stops getting run. A lot of my tooling work is making the loop between a change and its result shorter.

  3. 03

    A bug report should arrive with its cause.

    My triage bots sort product bugs from environment noise before a person looks, so the ticket that lands is worth reading.

  4. 04

    Own the whole loop.

    Runners, pipelines, reports and releases. Tests only matter if they run every time, where the team is looking.

Career

CV as a PDF

  1. Fintech startup

    Quality & AI automation engineer · name under NDA

    Aug 2026 to now

  2. Crypto wallet

    SDET, quality engineering · self-custody, used by millions

    May 2022 to Aug 2026

  3. Accenture

    QA engineer, test automation · enterprise systems

    2017 to May 2022

Toolbox

Testing
Playwright, Appium, WebdriverIO, Maestro, Selenium, Cypress, BrowserStack, Charles Proxy, Postman
CI and release
GitHub Actions, self-hosted Linux and macOS runners, test reporting, release sign-off
AI quality
LLM eval suites, AI failure triage, decision models, agent tooling
Code
TypeScript, JavaScript, Node, React, React Native, Next.js, Postgres (Supabase)
Platforms
iOS, Android, Electron desktop, browser extensions, web, REST APIs
Languages
Spanish (native), English (C2)