The interview consists of one phone screening and a virtual on-site consists of four rounds. I have given interview to AGI Data services, team. For phone screening I was asked a coding question of 2D binary search, ML breadth such as precision, recall, auc-roc score, evaluation metrics like perplexity. Virtual on-site consists of coding, ML depth, ML breadth and science application. Science application is about designing three evaluation metrics to select the high-quality dataset from the crowdsourced data. For coding I was asked to compute the chamfer distance efficiently.

Applied for Amazon AGI. After first round, it will go into full round of multiple interviews. Lots of modern LLM training technic questions. There are still some behavioral questions, but less than general Amazon roles.