Interview Questions

Dean Z. - "I will use below 4 pillars to rate and prioritize those 5 projects: Analyze the impact of those 5 projects, to our team and to the whole company maybe Find out if those 5 projects have any dependency between each other Clarify the timeline of each project Identity the status of resources to work on those 5 projects and also the status of their dependencies"See full answer

Anonymous Sparrow - "Clarifying question: why would Google consider sunsetting Gmail? Is there an opportunity to make a case against sunsetting or has the executive decision already been made and we're truly looking for a flawless sunset? If decision has already been made - why? Is this a global decision or country-specific? If the answer from interviewer is that the decision is open to debate, pivot case toward identifying the problem that causes leadership to even consider sunset. Then analyze and propose ways to"See full answer

Alex N. - "The adjusting context window size in LLM change trade off between reasoning capability, accuracy, computation cost. It influence how attention mechanist allocate resources across the input. Longer context window let it you input greater number of words and have more context to generate proper next token. However llms have lost in the middle issue. They remember the beginning of text and end of text but lost information located in the middle of long input. Another problem is Attention Dilution."See full answer

Ashley C. - "I follow a variation of the RICE framework when prioritizing how I ship product features. I start by looking at: Reach: Because the customer segmentation across our product portfolio is so similar, I tend to hold a lot of weight on product features that will maximize our customer reach with a minimal LOE. Impact: After establishing which customer segments will benefit from the product feature, I determine the urgency and estimated impact on each customer segment based on customer i"See full answer

Anchal V. - "supervised learning: model is trained on the labeled data unsupervised learning: no labels provided - model learns by finding patterns , structure and groupings in the data. Semi-supervised learning: use small set of labels to guide learning for the larger pool of unlabeled data. reinforcement learning: leans by interacting with students the environment, receives reward and penalties based on actions self supervised: no labelled data . The model makes its own practice problems by"See full answer

Venkata rakesh M. - "Here if we breakdown each dependency [A,B] , We need to check if there a cycle in Dependency Graph. If there is cycle installation is not possible, If there is no cycle installation is possible. Steps : 1: Build the graph 2: Perform DFS based Cycle Detection 3: Check each package if those packages have cycle or not."See full answer

Prateek G. - "I will first start with listing down the GOALS of Google Sheets as a product GOALS: Enable better collaboration amongst employees in an organisation Make edits parallel Share with co-workers so that the team can work seamlessly Increase user engagement, activity and make more users aware about the functionality and usefulness of Google Sheets Next, i will try and list the broader ACTIONS a user might take ACTIONS: Create a new Google Sheet Share with co-workers M"See full answer

Ramachandra N. - " def closest_palindrome(n: str) -> str: """ Finds the closest palindromic number to n (excluding itself). Assumptions: If two palindromes are equally close, return the smaller one. n is a positive integer represented as a string. Time Complexity: O(1) Space Complexity: O(1) """ length = len(n) num = int(n) Helper to build palindrome from a prefix def makepalindrome(prefix: int, isodd_length: bool) -> int: s = str(prefi"See full answer

Jacob S. - "Cloud computing is like going out to a restaurant instead of cooking dinner at home. When you cook at home, you have to do everything yourself. You need to use your own plates, pots, and pans. You have to know how many people are coming over and buy the right amount of ingredients. If more people join for dinner than you expected, you'll run out of food! But if fewer people show up, then you'll have leftovers that go to waste. And on top of it all, you have to do all the cooking, set up, and cl"See full answer

Michael A. - "First, I want to get a better sense of what is a DAU. What is considered “Active”? User sending message, reading message, opening app? All of the them, any activity within the Messenger App Second, it’s important to know the context of the change. What timeframe was this over? Let’s assume this was over the past 30 days. Ok, a follow up to this would be to check if this is a seasonal pattern? For example, there could be a correlation to Messenger usage and school starting. **This"See full answer

R S. - "One of the biggest risks I have taken was in a recent product, where we developed a new feature and released the feature externally without QA sign-off. Feature - Interactive Travel Itinerary Maps We had about ten days from the time we decided to include this new feature in the external launch. Usually, the biggest chunk of time in each release cycle was the sign-off from an external QA team. We fast-tracked the dev effort and completed the development of the feature in five days. Unfortunately,"See full answer

Pavani B. - "Request inputs/feedback on my response below with regards to this question and ways to improve and if it looks apt for this question. As a TPM at xxx, I was responsible for delivering a feature that would provide prescription (medications) alternatives to users in the Prescription Shopping Consumer Experience App. The project had a tight deadline as it was something that our account management team promised to a client who wanted it for their employees. My goal was to ensure smooth development"See full answer

Anonymous Stork - "For Xbox, the next-generation product I propose is a virtual reality headset. The rationale is threefold: Industry Trends: Competitors like Meta and Apple are investing heavily in VR/AR, signaling strong market potential. Xbox Legacy: Xbox has a history of pushing innovation in gaming hardware (e.g., Kinect), and VR represents the next frontier beyond traditional consoles. Strategic Need: With console sales flattening, Xbox must expand into new experiences that deepen engagement and different"See full answer

BriskD - "I watched a couple of videos like this, one of them by Exponent staff (I think). It was disappointing that the architecture diagram and the walkthrough was a general layered architecture that you can apply to any backend system. I was wondering if there are videos that can be considered a reference material to watch, learn and improve on the tech (sys design) skills, and not so much about the soft skills."See full answer

Anonymous I. - "Great question, first some clarifying questions: I assume that we're talking about meta disrupting the travel industry? Or generally When we say travel industry, this could include modes of transportation, scheduling transportation, what to do when you're there etc.? Are we focused on business travel? Leisure travel? I'm going to assume for the sake of scoping down that we're going to be focused on Leisure travel because business does not have as much pain, and on the flip-side o"See full answer

Anonymous Partridge - "We will have to use a second more powerful LLM Model to validate the answers. LLM as a judge"See full answer

R S. - "My main task was to improve our recommendation system. I worked closely with the business and data teams to understand how recommendations were made in the past. After discussions, I found out that the system had fixed recommendations and didn't consider customer preferences. Our goal was to identify potential customers better, boost conversion rates with personalized recommendations, and increase cross-selling of products. After analyzing a lot of data, we focused on customers with accounts an"See full answer

Sam E. - "average price per transaction would drop because we might not be considering currency exchange rates, which is much lower in India and Southeast Asia, when aggregating data globally. to define success, i'll first segment the data, separating existing markets to newer ones to check if that is the cause of the dropping average price. then i'll select a north star metric, which would be the number of completed transactions vs. attempted. i'll also track supporting metrics like usage frequency and"See full answer