An Amazon recruiter contacted me saying their speech and language team is interested in me. During the screen interview with the recruiter, she explained the whole interview process in Amazon including: 2 phone call interviews for everything listed on my resume, 1 coding question (Amazon sent me an email with a link for coding test) would be in either the first or the second phone interview. After the 2 phone interview, an on-site interview including a 30~40 min presentation and 5 1-1/1-2 interview will be held. My first phone interview asked me some speech recognition works I listed on my resume, also some machine learning questions, e.g., what is the difference between generative and discriminative models? what is SVM? why SVM need to use kernel? how SVM calculate the inner product in the high dimension space? what is PCA? what is LDA? what's the difference between PCA and LDA? how LDA measure the discriminant? No coding question was asked in this interview. After 3 days of the first phone interview, Amazon scheduled with me the second phone interview. The second phone interview which held in the next week was more formal and more speech technique oriented. There are two interviewer from Amazon speech team on the telephone interviewing me. It was a busy scheduled interview, which asked me a lot of details about my previous speech related research works, behavior questions, speech recognition technique details (e.g., how to build a speech recognition system from scratch? how to initialize the acoustic model at beginning? Why EM algorithm works? Explain Good-Turing and Kneser-Ney language model smoothing. How to do decoding (speech recognition)? and lots of other similar questions ) , and a coding question: given a dictionary and a character string, try to write a function segmenting the character string into a sequence of words listed in the dictionary. I didn't prepared well for this interview, so no follow up.