← Back

Ch.04 · How I keep up

Six things, done properly.

Manual and automation, API and database, performance and AI tooling. Each one opens to the tools behind it and one line of evidence, because a claim with nothing under it is just a word in a list.

Six disciplines side by side — manual and exploratory testing, automation, API and contracts, SQL and data, AI tooling — each listing its tools and one line of evidence.
Each discipline with the tools under it and the thing it produced. A skill with no evidence is a word in a list.
Ch.04What I do

Six things, done properly.

The part automation cannot do. Reading a spec, finding the case nobody wrote down, and reproducing it reliably.

  • Test design
  • Exploratory charters
  • Regression strategy
  • UAT
  • Defect triage

✓Designed and ran the test suites for 15+ delivered projects.

End-to-end suites that survive a redesign, on web and Android, wired into CI so they run without being asked.

  • Playwright
  • Selenium
  • Appium
  • Pytest
  • CI pipelines

✓Automated payout and instant-send workflows end to end; Appium suites for Android.

Contract and behaviour checks against the endpoints behind the UI, including the auth and idempotency edges the UI never shows you.

  • Postman
  • Newman
  • REST
  • Auth flows
  • Contract validation

✓Collection-driven API suites run in CI alongside the UI layer.

A green UI means nothing if the row is wrong. I check the database directly and reconcile it against what the API claimed.

  • SQL
  • DB verification
  • Reconciliation
  • Audit trails

✓Post-transaction DB checks on payment, payout and cashback flows.

Load and throughput on the flows that actually cost money: withdrawals, transactions, checkout under concurrency.

  • JMeter
  • Load profiling
  • Throughput analysis

✓JMeter scripts for withdrawal and high-volume transaction flows.

I use these every day for test design, failure triage, scripting, and reading a codebase I did not write. Then I used them to build Aegis-QA, the internal tool that holds all of this together.

  • Claude
  • Claude Code (CLI)
  • ChatGPT
  • Cursor
  • GitHub Copilot
  • Kimi 2.5
  • Emergent
  • Datadog AI

✓Used them to build Aegis-QA. Runs, API tests, SQL, Datadog and recorded scripts in one place.

—Toolkit

Automation

  • Playwright
  • Selenium
  • Appium
  • Pytest
  • Newman / Postman

Languages

  • Python
  • JavaScript
  • TypeScript
  • SQL

AI tooling

  • Claude
  • Claude Code (CLI)
  • ChatGPT
  • Cursor
  • GitHub Copilot
  • Kimi 2.5
  • Emergent
  • Datadog AI

Performance

  • JMeter
  • Load profiling
  • Transaction throughput

Domains

  • Crypto payments
  • Fintech
  • KYB / KYC
  • E-commerce
  • Games

Platform

  • FastAPI
  • Next.js
  • Docker
  • AWS EC2
  • CI pipelines

Next

Why any of it matters →

Six real defects, injected into a real page, caught by a real suite. Quality is an opinion until an assertion fails.