Ho presentato la mia candidatura online. La procedura ha richiesto 5 settimane. Ho sostenuto un colloquio presso Amazon nel mese di nov 2022
Colloquio
Online Assessment of two easy-medium Leetcode problems. Then second round, a one-on-one conversation with an Applied Scientist. He did not show up the first time and had to re-schedule. When he did show up, he asked me very basic ML questions, and several behavioral ones based on Amazon leadership principles. He did not ask about research/background. Seemed very disinterested.
Domande di colloquio [1]
Domanda 1
What is overfitting/ bias variance tradeoff/ regularization/ classification vs regression I'd love to know if there's anything one can do to get matched with better interviewers?
Ho presentato la mia candidatura online. Ho sostenuto un colloquio presso Amazon (Seattle, WA)
Colloquio
I went through two rounds of technical interviews. There was no coding at all; instead, I was asked about machine learning theory concepts as well as scenario-based applications and reasoning in ML.
Domande di colloquio [1]
Domanda 1
Leadership principles / bias variance trade off / gradient explosion
Ho presentato la mia candidatura online. La procedura ha richiesto 4 settimane. Ho sostenuto un colloquio presso Amazon (Sunnyvale, CA) nel mese di lug 2025
Colloquio
1st round: Assessment - Easy Data Structure Questions
2nd round: ML Depth - Very detailed, asked about transformers at a root level, Optimization questions on TensorFlowRT, T5 models
3rd round: ML Breadth - Basic ML questions like overfitting/underfitting.
Domande di colloquio [1]
Domanda 1
How does Multi-headed attention work?
What is serialization?
Ho presentato la mia candidatura online. La procedura ha richiesto 4 settimane. Ho sostenuto un colloquio presso Amazon (San Diego, CA) nel mese di gen 2025
Colloquio
Two back to back interviews focussed on breadth followed by depth covering pretty much the entire landscape of machine learning. They also have a lot of Leadership principles questions embedded into the interview. The challenging part is the breadth expectation.