Two-round interview. Each round lasts for 45 minutes. In every round, candidates will face a coding problem and then participate in a research talk. The coding problems are categorized as easy to medium in terms of difficulty, making them approachable for most candidates with a solid coding background. The research talks are designed to be relaxed and interactive, providing an opportunity for candidates to discuss and present their research work or interests in a conversational setting. This part is likely aimed at assessing communication skills and passion for research.

A smooth process. The interview consists of asking ML basic questions such as overfitting and dataset imbalance. Coding requires implementing convolution neural network. The interviewers are friendly. Two interviewers are involved.