Software Quality Assurance Course in Nepal, Manual Testing, Bug Reporting, API Testing & Playwright Automation

Saarathi's 8-week SQA track is designed to produce engineers who are genuinely strong at both. Three weeks of QA foundations, test design, manual testing and bug management, then a focused week of API and SQL testing, build the foundation. Three focused weeks of Playwright automation with Page Object Model, fixtures, and GitHub Actions CI/CD give you real automation depth, not a one-week overview.

What Real Entry-Level QA Jobs Actually Expect

• Manual testing depth: writing test cases, executing regression suites, exploratory testing with structured charters, and cross-browser and mobile validation.
• Bug reporting that developers can act on: clear reproduction steps, severity and priority classification, Jira workflow management, and defect lifecycle ownership.
• Agile QA participation: sprint ceremonies, acceptance criteria review, test sign-off, and release readiness communication.
• API testing with Postman: HTTP fundamentals, request/response validation, authentication flows, and chained collection runners.
• Database validation with SQL: verifying data state after API calls, boundary testing at the data layer, and reading backend state.
• Automation skills: Playwright with Page Object Model, cross-browser execution, and a basic CI/CD pipeline, not just clicking record in a tool.
• QA testing fundamentals: the vocabulary and test design techniques that appear in real interviews.
• A test-management tool alongside the bug tracker: Nepal job descriptions ask for both, so a real run cycle in TestRail, Zephyr or Qase is part of the course, plus the Excel test-case format Kathmandu teams still mail around.
• Integration and UAT work: both are named as day-job duties in live Kathmandu vacancies, so both are hands-on here rather than vocabulary.
• Security testing of web applications: Nepal mid-level QA posts ask for it by name, so you run the OWASP Top 10 checks a tester is expected to run and write the findings up as defects.
• Selenium, not just Playwright: Selenium is the first tool named in Nepal job ads, and hiring managers open your GitHub looking for a Selenium Page Object Model framework.

AI, Taught As Two Separate Skills

Employers hire for these separately, so we teach them separately. The first is AI as your tooling. Playwright bundles the MCP server and CLI since 1.62, so a coding agent can open a real browser and write a test against what it actually sees rather than hallucinating selectors, and the three Playwright Test Agents (planner, generator, healer) chain into a loop that produces coverage while you sleep. You use these with Copilot, Claude Code or Cursor from your first automation week, and you review every generated line, because a test you did not read is not a test you can defend. Ungoverned, these agents produce a hundred tests nobody trusts, so the graded skill is the review pass, not the prompt.

The second is testing AI itself, and that is where the market is short of people. Chatbots and LLM features never answer the same way twice, so exact-match assertions stop working and you need a different apparatus: a gold set of curated input and expected-output pairs, semantic and constraint-based assertions, LLM-as-a-judge scoring against a rubric you wrote, and for anything doing retrieval the RAG triad of faithfulness, answer relevance and context groundedness. Plus the failures that pull a chatbot from production: hallucinated numbers, prompt injection, leaked system prompts, and silent quality drift where a prompt tweak costs accuracy and no single test notices. One tool, done properly: promptfoo, open source, declarative YAML, one line in GitHub Actions.

This is directly employable in Nepal, not a future bet. F1Soft is advertising an AI Automation Engineer, Fusemachines is an AI company, and effectively every wallet, bank and telco in Kathmandu is shipping an AI feature somebody has to sign off. Almost no one applying for junior QA roles here can write an eval suite, so a public one on your GitHub is a real differentiator. What does not change is the judgement, deciding what is worth testing and whether a result is actually right, which is the part employers still hire humans for.

Built Against What Nepal Actually Advertises

This curriculum is checked against live Kathmandu vacancies rather than a generic global template. Entry-level posts here ask for SDLC, STLC and defect lifecycle, Postman API testing, SQL database validation, a bug tracker and a test-management tool, functional, integration, regression and UAT execution, and a contribution to test automation. Mid-level posts add security testing of web applications and release-management awareness. Nepal's tool stack names Selenium first, then Cypress and Playwright, with TestRail or Zephyr for test management, JMeter next to k6, and Jenkins next to GitHub Actions.

The biggest QA employers in Nepal are fintech (F1Soft, eSewa, Khalti, Fonepay), health and insurance outsourcing (Cotiviti, Deerwalk, Verisk), telecom (WorldLink) and product outsourcing (Leapfrog, Fusemachines). That is why payment and transaction testing is a first-class domain block here: transfer limits, OTP entry and expiry, double-submit idempotency, timeout mid-payment, failed and partial payments, refunds, and reconciling the screen against the database.

And the hardest part of a first QA job in Nepal is not the syllabus, it is the experience line. Entry posts routinely say 6 months to a year, and some posts titled "Junior QA Engineer" ask for 2 years. A certificate does not clear that bar; verifiable public work does. So the final week is built to produce evidence a hiring manager can check in ten minutes: two public automation repos, a Jira defect history, a test-management run report, a security findings write-up, three real defects filed against real public software, a 20-company target list and five tailored applications sent before you graduate.

Why This Curriculum is Structured the Way It Is

The manual testing phase comes first because automation that is not grounded in solid test design produces suites that pass but miss bugs. Understanding what to test and why is a separate skill from knowing how to automate it.

The API testing phase runs before automation because Playwright's request() API testing and Postman work cover the same HTTP knowledge from different angles. Learners who understand REST APIs manually write better automated API checks.

The Playwright automation phase runs for three weeks, not one, because Page Object Model, fixtures, cross-browser setup, visual regression, and GitHub Actions CI/CD each take real practice. One week of automation produces awareness, not skills.

What You Build and Take Away

• A complete test plan and test case library (60+ cases) for a real web application, covering functional, regression, and exploratory coverage.
• A Jira project with documented bug lifecycle history, defects filed, tracked, verified, and closed with proper evidence.
• A Postman API test collection with environment variables, chained requests, and response validation against a live API.
• SQL query examples for reading and verifying database state after API and UI actions.
• A Playwright automation framework on GitHub: Page Object Model structure, custom fixtures, cross-browser smoke suite, and a GitHub Actions CI workflow that runs on every pull request.
• A QA portfolio document, resume, and LinkedIn profile optimised for both manual and automation roles.
• Interview preparation and mock interview practice covering both manual and automation QA questions.