Proof of Work You Can Inspect
Each service links to an independent public project. Source code and
documentation are on GitHub, including what each project does not prove.
Payment Workflow Testing
Razorpay Payment Gateway QA & Automation
- What was tested
- A small Razorpay Test Mode checkout app: order creation, payment signature verification, amount integrity, webhook signature validation and idempotency, refund error handling, and the order state machine.
- Why it matters
- Payments cross the UI, the app, the gateway and asynchronous webhooks, and each can disagree with the others. Testing found two High-severity defects, both fixed and regression-tested.
- Tools
- Playwright Test (API and UI), Express, SQLite, GitHub Actions.
- What was automated
- 37 test definitions; 34 run in CI, 3 need live Test Mode credentials and are run manually.
- Limitations
- Sandbox only. Playwright drove a real Checkout payment in one documented session, but that session is not a committed, repeatable test. No webhook from Razorpay itself was observed arriving, and no successful live refund was completed.
API Testing
RESTful Booker API Automation
- What was tested
- The public RESTful Booker demo API across authentication, create, retrieve, full and partial update and delete, plus negative cases and multi-step workflows.
- Why it matters
- It shows how I structure API checks: status codes, response structure, data persistence across requests and rejection of invalid input.
- Tools
- Playwright Test, TypeScript, reusable API client classes, GitHub Actions.
- What was automated
- 47 test definitions in 9 suites, including 8 negative tests.
- Limitations
- A public practice API, not a client system. The last recorded CI run is 10 September 2026 and is a historical result.
Playwright Test Automation
Automation Exercise
- What was tested
- The Automation Exercise e-commerce demo: core customer journeys through the UI and selected REST endpoints.
- Why it matters
- It shows a maintainable UI and API regression suite built with page objects and run in CI, not a set of one-off scripts.
- Tools
- Playwright Test, Page Object Model, GitHub Actions.
- What was automated
- 26 test definitions: 16 API and 10 UI.
- Limitations
- A public demo site. The last recorded CI result is 29 August 2026 with retries enabled and is a historical result.
Manual & Functional Testing
Automation Exercise: the manual layer
- What was tested
- The same e-commerce demo, planned and tested manually first: a test plan, a manual test case log, exploratory testing using boundary-value analysis and state-transition heuristics, and a summary execution report.
- Why it matters
- Automation only covers what someone decided was worth checking. These documents show that decision step: scope, scenarios, execution results and a release assessment.
- Tools
- Manual and exploratory testing, documented in Google Docs; Playwright for the repeatable parts.
- Limitations
- A public demo site and an independent project. My employer project work is confidential and appears on this site only as generalized summaries.
These are independent, self-directed projects, not client or employer work.
Test counts are unique test definitions listed by each repository, not a measure of coverage.
CI run logs on GitHub require a sign-in.