Do know Amazon’s Leadership Principles. “Understand and be able to speak to how your experience aligns” with them. Also, be ready to apply “them to questions/hypotheticals that are posed during the hiring process.”
Do “brush up your machine learning and coding capabilities.” For example, solve coding questions on the internet in 15 minutes, listen to online courses on machine learning, or dive into textbooks.
Do “rest well the night before the on-site interview. If you are flying long hours and arrive in the afternoon, request to take another day in between.”
Do speak up. “Be a loud thinker. Discuss your thought process.” Also, have “a clear communication style.”
Do ask questions about the job, and show that you’re “interested in our efforts and [have] new ideas on how to improve them.”
Do be obsessed with the customer. “We would like to see candidates that put a considerable amount of weight on the customer experience when making decisions.”
Do wear business casual. “No one wears suits at Amazon.”
Don’t be vague. For example, “to avoid ambiguity the candidate can reply: I used SVM to build a classifier given 100K data points,” instead of “I used some technique to build a classifier with the given data points.”
Don’t be stubborn. You don’t want to demonstrate “an inability to think flexibly and open-mindedly” or be “missing hints from the interviewer and insisting on a particular approach.”