markqa

Decentralised liquidity-aggregation protocol for cross-chain swaps. Now 28+ chains, 766 liquidity sources, $8B+ swapped across 1.49M swaps for 194k users. Sole QA on a protocol that has since routed over $8B across 28 chains. Built the QA function from nothing: test plans, checklists, documentation and the release checks the team kept using. Wrote wallet-level automation with Synpress and Playwright — the genuinely hard part of Web3 testing, since MetaMask popups, signature prompts and mid-flow chain switches are invisible to a standard E2E driver. Reproduced failed on-chain transactions with Tenderly simulations, so a user's 'the swap failed' became a specific revert reason a developer could fix the same day. Ran user-issue triage across Zendesk, Discord and application logs, and handed developers the exact test data needed to reproduce each case rather than a description of it. Built Python and TypeScript tooling that compared our quoted prices against competing aggregators, which turned 'your rates are worse' from an argument into a measurement. Also contributed to pre-sale planning, protocol collaborations and community growth — useful range on a nine-person team.

Blockchain and fintech product engineering. Turned user stories into test conditions, then into plans, cases and checklists the team could run without me in the room. Ran API testing and integration validation across services, with frontend and backend log analysis to isolate causes rather than report symptoms. Built and ran performance and load tests with Locust and Python, giving the team numbers to design against instead of an opinion about capacity. Took part in incident analysis: found the cause, then wrote the guard so the same failure could not return unnoticed.

Ethereum scaling network, during the period it grew into one of the most-used L2 ecosystems. Team of 20, 4 QA. Extended an existing Java and Selenide automation framework with new scripts, and repaired failing ones rather than muting them to keep a dashboard green. Read automated run reports daily and drove every failure to a cause: test defect, product defect, or infrastructure. Getting that call wrong is how a real bug reaches users labelled as noise. Investigated customer-reported issues through application logs, reproduced them with steps a developer could follow, and worked with the customer directly to confirm what was actually broken.

All-in-one HR platform: payroll, benefits, compliance and analytics for US employers. Wrote E2E automation in JavaScript with Cypress against payroll and benefits flows, where a rounding error is a real person's paycheque, alongside manual and functional testing.