Interview Questions

Matt M. - "Given the dataset does not contain many labels, it implies we cannot directly use supervised learning. I would ask more about the type of dataset we are given. Is it images, text, etc? This may inform the types of transformations we do the dataset. I can see two approaches to training Given the labels we do have, we can find a method to generate labels for the other unlabeled data. This likely will introduce some error since they may not be true labels, but it at least allows processing the"See full answer

John C. - "Professional, congenial people"See full answer

R - "After further thinking on it, I take back the comments on phone-to-phone vs server-to-phone comments. Even if its p2p the smell data has to pass through a central messaging system. So we have to assume the technology to persist the smell also exists. So even (a) is possible. A perfume maker is capturing the smell and along with the perfume product attaching the smell data and posting on an e-commerce platform. Its similar to end-to-end encryption where data is encrypted in transit and at rest. I"See full answer

Anonymous Muskox - "Metrics which Youtube Consider before building a recommender system Number of likes on a video by user The watch time of a video by the user The video disklied by the user The video share by a user The video skipped or churn with 20-30 seconds. Depending on this Youtube build a recommender system. The video suggestion feature in youtube works based on the recommender system. It may use a hybrid of batch prediction and online prediction. So depending on the above metrics, the youtube p"See full answer

Exponent - "We've identified the problem as a Design a Product question. Use the following framework for tackling these types of questions: Ask Clarifying Questions Identify users, behaviors, and pain points State product goal Identify current solutions Brainstorm new solutions Evaluate solutions Measure success Summarize We'll go through each of these step by step. Ask Clarifying Questions The PM interview isn't about your ability to come up w"See full answer

Guru M. - "I have softened recruited people through LinkedIn. People from LinkedIn are more than happy to share their feedback on new products and features and to my surprise, I also saved a lot of company money on recruiting through services and offered Amazon vouchers to participants instead."See full answer

Anonymous Chickadee - "was there a specific scenario given for the interaction?"See full answer

Exponent - "This is a Technical question. It tests your ability to understand high level technical concepts. Even though your job won't have any coding involved, you'll still need to understand these concepts. Being able to cover all these topics with clarity communicates confidence in your interviewer. Unfortunately, there's no formula for technical questions, but some general tips are: Use analogies when you can Break your solution into clear, bite-size steps Don't be afraid to use examples to b"See full answer

Anmol R. - "The example given is wrong. The 2nd test case should have answer 3, as we can get to 6 by using 3 coins of denomination 2."See full answer

Exponent - "This is a Fermi problem — an estimation or approximation problem with limited information and back-of-the-envelope calculations. There's no right answer: interviewers want to understand how you think and how well you can explain your reasoning, rather than what you already know. Recall the formula for Fermi problems: Ask clarifying questions Catalog what you know Make equation(s) Think about edge cases to add to equation **Breakdown components of your equat"See full answer

Shibin P. - "Effective loss functions for computer vision models vary depending on the specific task, some commonly used loss functions for different tasks: Classification Cross-Entropy Loss:Used for multi-class classification tasks. Measures the difference between the predicted probability distribution and the true distribution. Binary Cross-Entropy Loss:Used for binary classification tasks. Evaluates the performance of a model by comparing predicted probabilities to the true binary labe"See full answer

Apurv M. - "Started with the clarifying questions. Then discussed the following with the panel: Goal Users Use Cases Features Priority Metrics to measure the success"See full answer