Code Reading — CLI Tool Feature Audit

Read an existing CLI-tool codebase against a written requirements list; identify missing features and propose optimisations. Asked as the first technical round in an Oracle Health AI IC4 onsite — not a traditional LeetCode round. Requirements The interviewer provides: A short written requirements document describing what a CLI tool is supposed to do. An existing implementation of that CLI tool. Task: Read the implementation against the requirements. Identify any features that are missing or incorrectly implemented. Propose specific optimisations (algorithmic, structural, or for maintainability). No green / red test signal — the interviewer is judging reasoning, code-reading speed, and how the candidate organises findings. Notes This is a code-reading / refactor round, not LeetCode. Pace yourself: 5-10 minutes reading the requirements, 10-15 minutes scanning the code in full before drilling into any one section. Take structured notes as you read: missing_features, bugs_present, optimisations, style_issues. Volunteer these in priority order at the end. Read the test surface first if tests are provided — they often reveal the interviewer's actual definition of "correct behaviour". When proposing optimisations, attach a concrete "before / after" sketch — "this nested loop is O(n²); a hashmap lookup makes it O(n)" is more persuasive than "this is slow." Be ready for the round to extend into a live coding follow-up: "now fix the most important issue you found." Pick one with the highest impact-to-effort ratio rather than the most interesting one. This round in OHAI is paired with a second round on applying a list of operations to an initial state (see apply-operations-command-pattern) — both deviate from traditional LeetCode. Preparation Practise code-reading on an unfamiliar small project (~500 lines) under a 20-minute timer; produce a structured findings list. Drill the "identify missing features against a written spec" workflow: read spec, read code, diff in your head. Brush up on common Java / Python performance gotchas (nested loops, boxed integers in tight loops, string concatenation in a loop) since proposed optimisations land best when phrased in language-idiomatic terms. Practise the "prioritise findings by impact" pitch — interviewers in this round score how you sequence your output, not just whether you spot everything.