Kakao Releases Solutions for Second Round of New Graduate Coding Test: A Battle of Trade-offs in Parallelism, Batching, and Exception Handling

Kakao released the problem intent and solutions for the second round of its new graduate online coding test, which took place on October 14. While the first round assessed basic implementation skills, the second round focused on evaluating the ability to acquire new technologies, design with trade-offs in mind, and monitor and improve. The problem involved implementing a crawler to collect image feature values using REST APIs and JSON parsing, encouraging iterative optimization through token validity limits and a real-time score leaderboard.

The scoring method weighted the number of correctly saved images and imposed penalties for missing data, undeleted data, incorrect data, and total query volume. Notably, candidates who applied parallel processing, batch processing, and exception handling all together scored around 250,000 points, whereas those who only implemented exception handling scored around 50,000 points. The passing score was set at 80,000 points, a level achievable without parallel processing if exception and batch handling were implemented well. The actual highest score was 235,969 points, and the average was 47,705 points. By language, Python (42.1%) and Java (35.1%) had the highest usage rates. Additionally, to ensure server stability, the request rate per token was limited to 50 requests per second, and CORS and JSON parsing issues that occurred during the test were addressed immediately to prevent any disadvantage to candidates.