TypeScript Refactor OA (HackerRank — fullstack)

Salesforce uses an existing-codebase exercise in full-stack loops: the detailed TypeScript form asks candidates to fix snapshot tests, add coverage, and refactor duplicated conditional logic while keeping tests green. A newer 90-minute OA includes a thinner Java Spring variant in which candidates modify the supplied project until its test cases pass. Requirements Input: a small TypeScript project with snapshot-style unit tests, some passing, some failing. Step 1: read the existing tests, identify and fix any broken ones. Step 2: add additional test coverage for un-tested branches. Step 3: refactor the production code. Typical refactor — replace nested if / else chains with a lookup table or rule-evaluator structure, eliminate duplicated logic, extract a clear extension point for new rules. All steps must keep tests green; the harness runs the suite continuously. Some interviewers also probe object-oriented / design-pattern justification ("which pattern did you apply?") during the closing chat. Notes This is labelled as a "system design" round on the loop schedule but the interview is a code-refactor exercise. If the interviewer agrees, you may push back early — but in practice opting out can be treated as a withdrawal. Plan to complete the refactor. Snapshot tests in Jest / Vitest: a fix usually means deleting the stale snapshot file and re-running with -u once the production behaviour is correct, not patching the snapshot string by hand. Common refactor targets: strategy pattern (one class per rule, dispatched by a map), table-driven evaluation (array of { matcher, action } rows), or extract-method + early-return. Hidden grader signals: clean commit-equivalent diffs (small, named, reversible), naming hygiene, and removing dead code as you go. Interviewers comment positively on candidates who narrate the refactor ("first I'll make this safe with a test, then change the structure"). Outcome feedback can be sparse: "told everything completed well, then rejected for not reaching senior bar with no specific gap explained" is the recurring complaint. Anchor expectations on the title (senior / staff / LMTS) — assume the bar is higher than the cosmetic difficulty of the task. Alternate canonical variant — Java Spring In a 90-minute full-stack OA, the second task is a small existing-code project selected from available language or framework choices. The candidate selected Java Spring and changed the code until the supplied tests passed. AI assistance is explicitly permitted for this project task, while the algorithm task in the same OA explicitly prohibits it. The exact Java interfaces and failing tests are not available yet. Preparation Practise a TypeScript refactor end-to-end: clone a small open-source TypeScript library, add an if-else chain, then refactor it into strategy / table-driven form under 45 minutes. Refresh Jest / Vitest snapshot semantics — how to add, update, and read snapshot files. Know expect(x).toMatchInlineSnapshot() vs toMatchSnapshot(). Be able to name and apply at least three classical design patterns (strategy, chain-of-responsibility, observer) and articulate which one matches the prompt. Walk through the GoF / Grokking-OOD strategy-pattern chapter — interviewers explicitly ask which pattern you chose and why. Practise entering a small Spring project cold, running its test suite, tracing one failure to production code, and making the smallest change that turns the suite green.