Amazon Interview Questions

Ilnur I. - "I can try to summarize their discussion as I remembered. Linear regression is one of the method to predict target (Y) using features (X). Formula for linear regression is a linear function of features. The aim is to choose coefficients (Teta) of the prediction function in such a way that the difference between target and prediction is least in average. This difference between target and prediction is called loss function. The form of this loss function could be dependent from the particular real"See full answer

Dhruv M. - "During my internship at Insighta Analytics, our team was planning to prioritize a marketing campaign focused on customer acquisition through social media ads. The marketing lead was confident this would yield the best ROI based on past success. I was tasked with supporting the strategy by analyzing customer engagement trends. But as I dove into the data, I started to question whether the current direction was the best use of our budget. I analyzed customer conversion data from the last two quart"See full answer

Nithesh S. - "First, let us start with the mission of Amazon: “We strive to offer our customers the lowest possible prices, the best available selection, and the utmost convenience.” Alignment of recommendations system with Amazon's mission: Suggesting recommendations to the users would be a part of this mission as it would allow users to buy things on their mind for the lowest cost in market (may or may not be true). They might get better prices if they order adjacent items together. For example, i"See full answer

Scott S. - "To recount an experience from my last project . We were at the initial stages of designing a new data platform to enhance our trading analytics capabilities. The conflict arose over the selection of a data processing framework. I was advocating for Apache Spark due to its scalability and performance benefits, especially for handling large volumes of derivatives data. Another senior team member preferred a different technology they had more experience with, which they believed would be easier to"See full answer

Ben G. - "In my time as a product designer, I’ve seen a transition of growth from something that is visually appealing to what is actually useful / functional / scalable. I think as designers, we are afraid to let things go such as the visual fidelity of design and how something could be useful but not necessarily have the flash of a more consumer based product. I think when we learn to let that go, we can see the scale of our impact and that we aren’t just people who make something look good/pretty but t"See full answer

Surbhi G. - "While serverless computing has its merits due to removing the need for server management and auto-scaling abilities, several challenges might arise when deploying machine learning models using this setup. Cold-start latency is a key issue where an Amazon Lambda function, for example, would take more time than usual to run after being idle. This could affect the performance of real-time ML applications negatively. A way to keep the functions warm might be required to avoid this latency. Moreover,"See full answer

Ali H. - "SQL databases are relational, NoSQL databases are non-relational. SQL databases use structured query language and have a predefined schema. NoSQL databases have dynamic schemas for unstructured data. SQL databases are vertically scalable, while NoSQL databases are horizontally scalable."See full answer

Anonymous Salmon - "Situation Action Result"See full answer

Anjali M. - "Colleague moved to a different role, so I decided to fill in and did outshine in showing results within the group."See full answer

Amir R. - "The key feature of echo is voice assistance. The demography for that audience has the following key behavior or use cases: Tech-savvy folks Use the voice assistant for their work (dictation service like writers, doctors, etc) Use for the driving assistance Use for the disabled Security and home automation."See full answer

Vaibhav C. - "To determine if a graph is not a tree, you can check for the following conditions: Presence of cycles: A graph is not a tree if it contains cycles. In a tree, there should be exactly one unique path between any two vertices. If you can find a cycle in the graph, it cannot be a tree. Insufficient number of edges: A tree with N vertices will have exactly N-1 edges. If the graph has fewer or more than N-1 edges, then it is not a tree. Disconnected components: A tree is a connected graph, m"See full answer

Ds S. - "Initially I asked clarifying questions like whether the tree can be empty or not and asked the interviewer to explain what is meant by left view and the explanation for the sample inputs. Then I came up with the level order traversal approach where we visit each level in the binary tree at once using a queue and at each level print the value of the first node. Interviewer seemed satisfied with the approach and asked me to code it up. Finally gave the time and space complexity of the solution."See full answer

Ben G. - "Situation - A time I saw a peer struggling was while I was at Google working in Mountain View. A fellow colleague of mine was consistently not meeting standards in his performance reviews and wasn’t sure why. He was upset because Google was his dream company and he wasn’t sure how to improve based on his feedback. Task - I was tasked to design a based VUI framework for our discovery space on smart displays. This colleague was acting as the VUI lead for the project while I acted as t"See full answer