Kakao Releases Explanations for 2019 Blind Open Recruitment Second Coding Test Problems

Kakao released explanations for the problems in the second offline coding test of the 2019 blind open recruitment, explicitly stating that system design capability was the core evaluation element of this year's exam. Unlike last year, the focus was on algorithm design and modularization skills to efficiently control multiple elevators, going beyond simple implementation.

Problem Structure and Requirements

The exam was a task to implement a system controlling elevators from 1 to 4, requiring candidates to handle passenger requests (Call) and manage elevator status (Status) via REST API communication with the server. Candidates had to implement optimal control logic tailored to three building environments: Apeach Mansion (5 floors, 6 requests), Jay G. Building (25 floors, 200 requests), and Ryan Tower (25 floors, 500 requests).

Evaluation Criteria and Algorithm Performance

The evaluation criterion was the final timestamp required to transport all passengers to their destinations. When applying the common Collective Control (LOOK) algorithm, it was found that approximately 972 timestamps were required for Jay G. Building and approximately 2000 for Ryan Tower. On the other hand, higher scores could be achieved through various optimization strategies such as utilizing prior information (pre-recognizing destination floors), consolidated boarding after waiting, and space utilization.

Statistics and Release

In language usage statistics, Python accounted for an overwhelmingly high proportion, followed by Java and Node.js. The first successful submission times for each building were recorded as 14:12, 14:40, and 14:53. Kakao released the elevator simulation server used for the test, allowing anyone to run it.