Case studies

Projects & testing outcomes

Four products, four different kinds of risk. These write-ups cover what made each one hard to test and what I actually did about it — not just a list of features.

Hangar Direct

Aviation web platform

Senior SQA Engineer & Project Coordinator

A booking and management platform for aircraft hangar space and maintenance scheduling.

The challenge

Multi-role permissions, real-time availability, and money-adjacent booking flows meant a single missed edge case could double-book a hangar or expose another operator's data.

Testing strategy

  • Mapped every user role against every action to build an explicit permission matrix, then tested each cell rather than sampling.
  • Validated all API endpoints in Postman, including negative cases: expired sessions, tampered IDs, and out-of-range dates.
  • Checked transactional integrity directly in PostgreSQL after each booking flow to confirm no partial writes survived a failure.
  • Coordinated fixes and re-tests through Jira, keeping a clean audit trail from report to verification.

Outcome

Permission and booking defects were found and closed before launch rather than in production, and every reported ticket was re-verified against the fix before release sign-off.

  • Next.js
  • PostgreSQL
  • Postman
  • Jira

RideSkipper

iOS & Android chauffeur booking

Senior SQA Engineer

A cross-platform app connecting passengers with professional chauffeurs, with live tracking and in-app payment.

The challenge

Location tracking and payment both fail in ways an emulator never shows — dropped GPS signal, backgrounded apps, interrupted payment callbacks — and any of them loses a real customer's money or ride.

Testing strategy

  • Tested on physical Android and iOS devices, not just simulators, to catch platform-specific behaviour.
  • Simulated GPS movement and signal loss to verify tracking recovered instead of silently stalling.
  • Exercised payment flows against gateway test modes, including declines, timeouts, and duplicate submissions.
  • Verified push notification delivery across foreground, background, and terminated app states.

Outcome

The app reached App Store and Play Store release with booking, tracking, and payment flows validated on real devices across both platforms.

  • React Native
  • MongoDB
  • Postman
  • Mobile device testing

QS ERP

Real-estate ERP

SQA Engineer

An enterprise ERP for property management, sales tracking, and customer relationships.

The challenge

ERP modules share data, so a defect in one screen surfaces as corrupted totals three modules away — and the UI often looks correct while the underlying records are not.

Testing strategy

  • Traced each business workflow end to end across modules instead of testing screens in isolation.
  • Wrote SQL queries against MS SQL Server to verify what was actually persisted after each operation.
  • Built a regression suite around the highest-risk financial calculations and re-ran it every release.
  • Ran UAT sessions with client stakeholders, translating their business language into reproducible test cases.

Outcome

Data-integrity defects that were invisible in the interface were caught at the database layer before they reached client reporting.

  • MS SQL Server
  • PowerBuilder
  • Jira

Carpet Market

eCommerce

SQA Engineer

An online carpet retailer with catalogue, cart, and order management.

The challenge

In eCommerce every checkout defect is lost revenue, and the failure modes cluster in exactly the places that are tedious to test: cart state, stock counts, and price totals.

Testing strategy

  • Tested the full purchase path — browse, filter, cart, checkout, confirmation — including abandoned and resumed carts.
  • Verified totals, discounts, and stock decrements against the database rather than trusting the displayed value.
  • Ran cross-browser checks on the catalogue and checkout to catch layout and validation differences.
  • Confirmed order records and status transitions in the admin panel matched what the customer saw.

Outcome

Checkout and stock-handling defects were resolved before launch, and the catalogue rendered consistently across the supported browsers.

  • PostgreSQL
  • Regression testing
  • Cross-browser testing

Have a release that needs testing properly?

I take on contract QA work — a single release, an ongoing regression safety net, or setting up a QA process for a team that doesn't have one yet.