ML System Design: RecSys interview questions

Design a recommendation system for an e-commerce platform that can handle 100,000 concurrent users and provides a response time of less than 200 ms per query. You have a budget of $50,000 for infrastructure and a team of 5 people. What key components would you utilize?

Design a recommendation system for a music streaming service that must handle 10 million users with 1000 QPS and provide personalized playlists. What architecture would you propose to ensure scalability and low latency?

Design a news feed ranking system that must update in real-time for 5 million users with a latency limit of no more than 500 ms. How will you handle new articles and account for user preferences?

Design a 'similar items' recommendation system for an e-commerce platform that must handle 50 million products and 20 million users. Response time should be no more than 100 ms. What algorithms and methods would you apply?

Design an ad targeting system that must provide high relevance for 10 million users in real-time with a query rate of 5000 QPS. How will you incorporate user data and their actions?

Design a 'people also bought' system for an e-commerce platform that must handle 30 million users and 100 million products, with latency of no more than 300 ms. What data and metrics will you use to build the model?

Design a personalized email marketing system for 5 million users that must be tailored to user preferences and achieve high open rates. How will you manage user data and their behaviors?

Design a recommendation system for a video streaming service that must handle 15 million users and ensure response times of less than 200 ms per query. What approaches will you use to implement this project?

Design a content recommendation system for a news website that must account for 10 million unique users and their preferences. What strategy will you choose for updating the model?

Design a 'similar videos' system for a video service that must handle 20 million users and ensure response times of less than 150 ms per query. What factors will you consider when building the recommendation model?

Design a recommendation system for an e-commerce platform that will handle 1 million users and 10,000 queries per second. The system should provide personalized product recommendations in response to user queries within 100 ms. Consider a limited budget of $50,000 for data storage.

Build a recommendation system for a music streaming service that will handle 5 million users and support 5,000 queries per second. The system should return results within 200 ms and consider user preferences based on their listening history. Discuss the trade-offs between accuracy and response time.

Design a news feed ranking system for a social network that will handle 2 million active users with 20,000 queries per second. The system should take into account user interactions with content and return results within 150 ms. You will need to choose an appropriate ranking algorithm.

Design an ad targeting system for an online store that will handle 1 million users and 15,000 queries per second. The system should return suitable ads within 80 ms, based on user behavior and preferences. Discuss privacy concerns related to data usage.

Design a 'Similar Items' recommendation system for e-commerce that must handle 3 million users with 8,000 queries per second. The system should return results within 150 ms and utilize both product and user data. Discuss architectural decisions for scalability.

Design a personalized email system for an online store that will handle 500,000 users and send 10,000 emails per second. The system should generate personalized content based on user behavior analysis and create emails within 300 ms. Discuss trade-offs between personalization and performance.

Design a recommendation system for a video platform that will handle 10 million users and 25,000 queries per second. The system should return results within 200 ms and consider user preferences and current trends. Discuss how to achieve a balance between relevance and diversity of recommendations.

Design a system that recommends news sections to users based on their interests. The system must handle 1 million users and 12,000 queries per second, returning recommendations within 100 ms. Discuss how machine learning can be used to enhance the quality of recommendations.

Design a 'People Also Bought' system for an online store that will handle 2 million users and 5,000 queries per second. The system should return results within 150 ms and consider overall purchase history. Discuss how graphs can be utilized to improve recommendations.

Design a recommendation system for a streaming video service that will handle 8 million users and 30,000 queries per second. The system should provide recommendations within 250 ms and consider various factors such as watch time and ratings. Discuss approaches to processing large volumes of data in real-time.