Pergunta de entrevista da empresa Planck

1. Service Implementation with Smart Integration Tests Goal: Implement a simple service that greets a user by name, with robust integration test coverage. Requirements: A REST endpoint that returns "Hello, !" Name can be received either as: Path variable (e.g., /hello/John) Or query parameter (e.g., /hello?name=John) If the name is missing, return an appropriate error (400 Bad Request) Write integration tests that: Generate dynamic names (e.g., using UUID/random names) Test both input methods (path/query) Ensure the response is correct even when name is missing 2. Add Time-Based Blocking Logic (Smartly Testable) Goal: Add a rule to block the service between 14:00 and 16:00, in a way that can be mocked and tested reliably. ✅ Requirements: When the request comes between 14:00–16:00 → return 503 Service Unavailable Do NOT use LocalTime.now() directly inside logic. Instead: Abstract time-checking logic into a TimeGatekeeper interface with a method like boolean isAllowedNow() Implement a default version that checks current time In tests: Mock the TimeGatekeeper to simulate allowed/blocked times Ensure greeting still works in tests (time-blocking shouldn’t interfere with unrelated test logic) and you shold not mock the clock ✅ Also: Write a unit test for TimeGatekeeper itself Integration tests should pass even if current system time is in blocked range 3. System Design Question: OCR Over Millions of Images in S3 Scenario: You have: A large S3 bucket s3://images-a with millions of images (4–5MB each) A single machine with limited memory, CPU, and disk Goal: Run OCR on all images and save results to a second bucket s3://images-b Constraints: Maximize throughput, support failure recovery, and enable monitoring