I build QA where there isn't any, and fix it where it stopped scaling. Test strategy, exploratory and regression testing, mobile device coverage, and API validation, with defect reports your developers can reproduce on the first read.
What I do
Risk-based coverage decisions, a written test plan, and a test case library that survives after I leave. For teams with no QA process, or one that stopped scaling.
Structured exploratory sessions that find what scripted suites walk past, plus a regression pass you can run against every release. Defects arrive reproducible, with screenshots or video.
Real-device validation across mobile devices and desktop browsers, covering OS versions, manufacturers and screen sizes. I build the matrix, justify what gets cut, and test the cut list on the riskiest builds.
Locust profiles built around the worst realistic hour, not average traffic. Confirms headroom is reserved ahead of a spike rather than scaled into it, and that nothing degrades when it matters most.
Validation of AI-driven features against requirements, so outputs are accurate and reliable before they reach users. I also use AI to speed up routine test work, and can set up local LLM infrastructure so your test data never leaves your environment and costs stay predictable.
REST testing in Postman with assertions and environments, backend verification in PostgreSQL, MySQL, SQL Server and DynamoDB, and third-party integration testing where the contract is not yours to control.
Selected work
A caller ID software product built for multiple carriers and pre-installed on their handsets relied on manual regression passes, so I built automation to cut execution time and free up hours for deeper testing.
A full-service software development firm was shipping web, mobile and custom software for clients with no standardized QA process. Several of those projects were high-stakes systems where a defect is measured in hours, not tickets.
A regulated online investment platform needed test infrastructure and a defect process that would hold up under compliance scrutiny, without slowing delivery.
Deliverables
Sample deliverables, so you can see the quality of the work before you hire. Nothing here is client confidential.
Test strategy
Scope, out of scope, a risk register, environments, and entry and exit criteria for a subscription billing app. Shows how I decide what not to test.
Test case suite
Twelve cases across login, payment, plan changes and webhooks. Preconditions stated, expected results specific enough to fail on.
Defect reports
Three reports at blocker, major and minor severity. Reproducible on first read, which is the entire job of a defect report.
Device coverage matrix
A tiered device matrix with the reasoning behind each entry, and an explicit cut list of what is not covered.
API test collection
Postman collection structure, contract-level assertions, environments, and a Newman run wired for CI.
Automation framework
PyTest and Playwright: page objects, fixtures, failure evidence, and the rules that keep a suite maintainable.
Background
I’ve built QA from scratch four times, covering process, tooling and infrastructure, including at a regulated fintech platform under launch pressure and at two development agencies running up to a dozen client projects at once. I also spent eight years testing Android and iOS software, much of it shipped through carrier and OEM partners, where a rejected build costs a release cycle.
I write automation in Python, PyTest and Playwright where it earns its keep, and I say so plainly when it doesn’t. Most of the value I add is upstream: deciding what to test, in what order, and what the business risk actually is. That judgment comes from years of watching where software breaks, and it’s the one thing no tool can supply. I use AI to speed up the routine work, such as test data, scaffolding and triage, so more of your budget goes on real testing, and every result is still checked by me.
Some of that work carries real consequences. I’ve tested multi-tenant public safety systems where isolation between agencies is a functional requirement, AI-assisted workflows that must never decide anything on their own, and platforms that have to hold under the worst hour they will ever see rather than an average one.
Available full time. Remote — US.
Contact
Send me a staging link and what you're worried about. I'll tell you where I'd start.