Interview Questions

Review this list of 4,065 interview questions and answers verified by hiring managers and candidates.
  • +2

    "(This is debugging, then trade off question) WAU and email open rates can be influencing each other, especially an email opened and then clicked could lead to an active user, but it doesn't necessarily mean that they have correlations or be the only reason causing changes. The approach So we definitely need to look at each cause of Why MAU goes up, and why email notification open rates go down, then develop some hypothesis on for proper cause, I'd love to gather some info to narr"

    Scarlett S. - "(This is debugging, then trade off question) WAU and email open rates can be influencing each other, especially an email opened and then clicked could lead to an active user, but it doesn't necessarily mean that they have correlations or be the only reason causing changes. The approach So we definitely need to look at each cause of Why MAU goes up, and why email notification open rates go down, then develop some hypothesis on for proper cause, I'd love to gather some info to narr"See full answer

    Execution
    Analytical
  • OpenAI logoAsked at OpenAI 

    "Reinforcement Learning is a type of machine learning where an agent learns to make decisions by trying out different actions and receiving rewards or penalties in return. The goal is to learn, over time, which actions yield the highest rewards. There are three core components in RL: The agent — the learner or decision-maker (e.g., an algorithm or robot), The environment — everything the agent interacts with, Actions and rewards — the agent takes actions, and the environmen"

    Constantin P. - "Reinforcement Learning is a type of machine learning where an agent learns to make decisions by trying out different actions and receiving rewards or penalties in return. The goal is to learn, over time, which actions yield the highest rewards. There are three core components in RL: The agent — the learner or decision-maker (e.g., an algorithm or robot), The environment — everything the agent interacts with, Actions and rewards — the agent takes actions, and the environmen"See full answer

    Machine Learning Engineer
    Concept
    +1 more
  • Google logoAsked at Google 

    "I'd like to clarify the question here for better understanding. Are we targeting any location? Assuming WW Do we have any demographics of the users? Assuming overall users Did we see any drop recently in TV watching users metrics for which we want to improve the watch time on TVs? No, in general we want to improve this area. Are we looking to improve the overall Netflix watch time on TV, not to acquire new users? Correct. As Netflix is offering content of so many genres of con"

    Vijendar K. - "I'd like to clarify the question here for better understanding. Are we targeting any location? Assuming WW Do we have any demographics of the users? Assuming overall users Did we see any drop recently in TV watching users metrics for which we want to improve the watch time on TVs? No, in general we want to improve this area. Are we looking to improve the overall Netflix watch time on TV, not to acquire new users? Correct. As Netflix is offering content of so many genres of con"See full answer

    Product Manager
    Product Design
  • "👇 Your feedback is very much appreciated 👇 Reduce scope of product for the initial launch. Streamline the PRD to an actual MVP and use the current PRD (that takes 6 months to build) for the V1"

    Julien C. - "👇 Your feedback is very much appreciated 👇 Reduce scope of product for the initial launch. Streamline the PRD to an actual MVP and use the current PRD (that takes 6 months to build) for the V1"See full answer

    Product Manager
    Execution
    +1 more
  • Meta (Facebook) logoAsked at Meta (Facebook) 

    "Functional requirement's: partial search while searching for users, products any keywords in the search. additional keywords in the filter Black listed words in the search. Non functional requirements: low latency, search through 2 Billion records recent search should be cached. Design: high reads, we should have caching enabled over the primary db storages. caching cluster can be added when the search load increases. read ahead. - check in cache (periodic cache refresh), lfu, lru "

    Sandeep Y. - "Functional requirement's: partial search while searching for users, products any keywords in the search. additional keywords in the filter Black listed words in the search. Non functional requirements: low latency, search through 2 Billion records recent search should be cached. Design: high reads, we should have caching enabled over the primary db storages. caching cluster can be added when the search load increases. read ahead. - check in cache (periodic cache refresh), lfu, lru "See full answer

    Machine Learning Engineer
    System Design
  • 🧠 Want an expert answer to a question? Saving questions lets us know what content to make next.

  • Product Manager
    Analytical
  • Google logoAsked at Google 
    +1

    "Goal of the engine : 1. Recommend the "right" videos 2. Make users watch the videos continuously The design consists of 4 different components : Video database (corpus): Consists of million of videos. Recommendation Engine : Based on User history (watch and search history, when enabled) coupled with user context (country, time of the day), the millions of videos are filtered to hundreds of videos and passed through the recommendation engine Ranking : Apart from user history and contex"

    Anjaly J. - "Goal of the engine : 1. Recommend the "right" videos 2. Make users watch the videos continuously The design consists of 4 different components : Video database (corpus): Consists of million of videos. Recommendation Engine : Based on User history (watch and search history, when enabled) coupled with user context (country, time of the day), the millions of videos are filtered to hundreds of videos and passed through the recommendation engine Ranking : Apart from user history and contex"See full answer

    Technical
  • Adobe logoAsked at Adobe 
    Software Engineer
    Data Structures & Algorithms
    +4 more
  • "The central limit theorem tells us that as we repeat the sampling process of an statistic (n > 30), the sampling distribution of that statistic approximates the normal distribution regardless of the original population's distribution. This theorem is useful because it allows us to apply inference with tools that assume normality like t-test, ANOVA, calculate p-values hypothesis testing or regression analysis, calculate confidence intervals, etc."

    Lucas G. - "The central limit theorem tells us that as we repeat the sampling process of an statistic (n > 30), the sampling distribution of that statistic approximates the normal distribution regardless of the original population's distribution. This theorem is useful because it allows us to apply inference with tools that assume normality like t-test, ANOVA, calculate p-values hypothesis testing or regression analysis, calculate confidence intervals, etc."See full answer

    Statistics & Experimentation
  • Amazon logoAsked at Amazon 

    "Functional Requirement Monitor health, metrics Alert in case of failure/anomaly Visualize the live health Analyse machines on periodic basis Non Functional Should not exert load on machines low latency Highly scalable Logs/Metrics Gathering push - machine gather and send to system and low priority background thread along with batching pull - heart beat check (for offline machines) Processing Real time streaming using Kafka/kinesis + Flink TimeSeries database for stor"

    Sourabh G. - "Functional Requirement Monitor health, metrics Alert in case of failure/anomaly Visualize the live health Analyse machines on periodic basis Non Functional Should not exert load on machines low latency Highly scalable Logs/Metrics Gathering push - machine gather and send to system and low priority background thread along with batching pull - heart beat check (for offline machines) Processing Real time streaming using Kafka/kinesis + Flink TimeSeries database for stor"See full answer

    Engineering Manager
    System Design
    +2 more
  • Google logoAsked at Google 

    "As a YouTube PM, there are several factors that I would consider when evaluating the idea of developing a tool for content creators to generate ideas with scripts automatically added. First, I would assess the potential demand for such a tool among YouTube content creators. This would involve researching the needs and challenges of content creators, as well as gathering feedback from content creators on the value and usefulness of the tool. This would help me understand whether there is a strong"

    Anonymous Flamingo - "As a YouTube PM, there are several factors that I would consider when evaluating the idea of developing a tool for content creators to generate ideas with scripts automatically added. First, I would assess the potential demand for such a tool among YouTube content creators. This would involve researching the needs and challenges of content creators, as well as gathering feedback from content creators on the value and usefulness of the tool. This would help me understand whether there is a strong"See full answer

    Product Manager
    Product Strategy
    +1 more
  • Zillow logoAsked at Zillow 
    Video answer for 'Design Zillow.'

    "What about sharding by Real Estate companies that are listing the homes? Because the homes can span zip codes. Then further shard by zip codes?"

    Adrian V. - "What about sharding by Real Estate companies that are listing the homes? Because the homes can span zip codes. Then further shard by zip codes?"See full answer

    Engineering Manager
    System Design
    +1 more
  • Amazon logoAsked at Amazon 
    Video answer for 'How do you split a machine learning dataset for training, evaluation, and testing?'

    "It depends on the size of the dataset. You want enough samples in both the testing, training and evaluation sets. If there is enough data, 70/20/10 is a good split"

    Jasmine Y. - "It depends on the size of the dataset. You want enough samples in both the testing, training and evaluation sets. If there is enough data, 70/20/10 is a good split"See full answer

    Coding
    Data Structures & Algorithms
  • Airbnb logoAsked at Airbnb 
    +1

    "Each team member is different, understand their needs, strengths, weakness areas. Motivate accordingly. Frequent 1:1 and help in their career growth. According to me micromanagement never wins, so don't even try it. Inform team high level picture in case if some work is not challenging."

    BePostive - "Each team member is different, understand their needs, strengths, weakness areas. Motivate accordingly. Frequent 1:1 and help in their career growth. According to me micromanagement never wins, so don't even try it. Inform team high level picture in case if some work is not challenging."See full answer

    Engineering Manager
    Behavioral
  • "Context: As a people manager, my biggest thing is to foster psychological safety for my team members and the way I do that: building rapport between me and my direct reports, weekly team meetings I ensure all voices are heard, have people share about themselves and encourage them to get to know one another. Team members had a good understanding of each other’s working style. One of the recent projects my team was responsible for building out the espresso test framework on android. This was"

    Shawn S. - "Context: As a people manager, my biggest thing is to foster psychological safety for my team members and the way I do that: building rapport between me and my direct reports, weekly team meetings I ensure all voices are heard, have people share about themselves and encourage them to get to know one another. Team members had a good understanding of each other’s working style. One of the recent projects my team was responsible for building out the espresso test framework on android. This was"See full answer

    Engineering Manager
    Execution
    +2 more
  • "% of content watched by clicking on recommended content Content watched by clicking on recommended content : Content watched through search"

    Medha A. - "% of content watched by clicking on recommended content Content watched by clicking on recommended content : Content watched through search"See full answer

    Product Manager
  • Microsoft logoAsked at Microsoft 

    "Assumptions, company, industry, external, customers, problems, solutions, metrics Company is microsoft, its vision is to help people and organization to achieve their full potential by increasing productivity . Task management - there are many products like notes, google task etc which exist but no one is market leader. A user had a list of tasks which she wants to complete by Y deadline and their is prirotization among these and could be categorized Task - start day, end day, estimated time to"

    Megha V. - "Assumptions, company, industry, external, customers, problems, solutions, metrics Company is microsoft, its vision is to help people and organization to achieve their full potential by increasing productivity . Task management - there are many products like notes, google task etc which exist but no one is market leader. A user had a list of tasks which she wants to complete by Y deadline and their is prirotization among these and could be categorized Task - start day, end day, estimated time to"See full answer

    Product Manager
    Product Design
  • Adobe logoAsked at Adobe 
    Video answer for 'Generate Parentheses'
    +5

    "class Solution: def generateParenthesis(self, n: int) -> List[str]: stack = [] res = [] def backtrack(openN, closedN): if openN == closedN == n: res.append("".join(stack)) return if openN < n: stack.append("(") backtrack(openN + 1, closedN) stack.pop() if closedN < openN: stack.append(")") backtrack(openN, clo"

    Anonymous Roadrunner - "class Solution: def generateParenthesis(self, n: int) -> List[str]: stack = [] res = [] def backtrack(openN, closedN): if openN == closedN == n: res.append("".join(stack)) return if openN < n: stack.append("(") backtrack(openN + 1, closedN) stack.pop() if closedN < openN: stack.append(")") backtrack(openN, clo"See full answer

    Software Engineer
    Data Structures & Algorithms
    +3 more
Showing 1341-1360 of 4065