ML & GenAI system design interview preparation.
Go from product requirements to data, modeling, serving, evaluation, reliability, and iteration. Includes classic ML systems plus modern RAG and LLM architecture.
High-signal system design questions
Explain Transformer architecture and each major component.GenAI & LLM · medium · Evidence 88/100Design a content moderation ML systemML System Design · hard · Evidence 77/100Improve Instagram comment rankingML System Design · hard · Evidence 77/100Design an ML system using traffic dataML System Design · hard · Evidence 77/100Design an end-to-end image-and-text pipeline to predict document aspect ratio.ML System Design · hard · Evidence 77/100Design a color suggestion system.ML System Design · hard · Evidence 77/100Design entity matching across different Alexa skills.ML System Design · hard · Evidence 77/100How would you handle cold start for new ads?ML System Design · medium · Evidence 77/100Rank nearby places beyond distanceML System Design · hard · Evidence 77/100Design local search rankingML System Design · medium · Evidence 77/100Design Nearby Places RecommendationsML System Design · hard · Evidence 77/100Design a cold-start recommenderML System Design · hard · Evidence 77/100Predict whether an old post is good for MemoriesML System Design · medium · Evidence 77/100Predict the language of Instagram postsML System Design · medium · Evidence 77/100Design a Traffic-Data ML SystemML System Design · hard · Evidence 77/100Design News Feed rankingML System Design · medium · Evidence 77/100Explain core NLP and deep-learning concepts used in an Alexa system.GenAI & LLM · medium · Evidence 77/100Design a recommendation system for a product scenario.ML System Design · hard · Evidence 75/100Design a system to detect copyright violations on YouTubeML System Design · hard · Evidence 75/100How would you forecast daily grocery inventory for the next week?ML System Design · medium · Evidence 75/100How would you reason about distributed-systems trade-offs in an ML product?ML System Design · hard · Evidence 75/100Design a small polite LLM that runs on a phoneGenAI & LLM · hard · Evidence 75/100How can Transformers be used in recommendation systems?GenAI & LLM · hard · Evidence 75/100Design a RAG system that generates presentation slides from a user prompt.GenAI & LLM · hard · Evidence 75/100Explain how RAG works end to end.GenAI & LLM · medium · Evidence 75/100Design a RAG system end to end.GenAI & LLM · hard · Evidence 75/100Explain the architecture of attention.GenAI & LLM · medium · Evidence 75/100Design a short-video recommendation system similar to Reels.ML System Design · hard · Evidence 70/100Why use RAG instead of fine-tuning an LLM?GenAI & LLM · medium · Evidence 66/100Design Gun-Violence Video ModerationML System Design · hard · Evidence 66/100Design a Pinterest Recommendation SystemML System Design · hard · Evidence 66/100Design a Sales-Based Ranking Data StructureML System Design · medium · Evidence 66/100Design Text-to-Image SearchML System Design · hard · Evidence 66/100How would you investigate model drift in production?ML System Design · medium · Evidence 58/100How would you scope down an ambiguous ML system-design problem before modeling?ML System Design · medium · Evidence 41/100Design a Q&A BotML System Design · hard · Evidence 41/100Design People You May KnowML System Design · hard · Evidence 41/100How would you formulate an ML problem from a product goal?ML System Design · medium · Evidence 41/100Design a LinkedIn Classification SystemML System Design · hard · Evidence 41/100Design a Places Recommendation System for MapsML System Design · hard · Evidence 41/100How would you turn a research model into a production ML system?ML System Design · hard · Evidence 41/100How would you model structured and unstructured signals from the same document?ML System Design · hard · Evidence 41/100How would you reason about product sense for an ML ranking feature?ML System Design · medium · Evidence 41/100How would you generate candidates for a brand-new user before personalization data exists?ML System Design · medium · Evidence 41/100Design a Recommendation System at High LevelML System Design · hard · Evidence 41/100How would you monitor distribution drift in a multimodal model?ML System Design · hard · Evidence 41/100How would you avoid feedback loops in a professional recommendation system?ML System Design · hard · Evidence 41/100Design Recommendation + Ranking for a Hiring-Manager RoundML System Design · hard · Evidence 41/100How would you profile GPU inference latency end to end?ML System Design · hard · Evidence 41/100How do throughput and tail latency trade off in model serving?ML System Design · hard · Evidence 41/100Design Global Search for Ephemeral Status PostsML System Design · hard · Evidence 41/100How would you handle sparse implicit feedback in a cold-start recommender?ML System Design · hard · Evidence 41/100Design a Bot-Post Detection PipelineML System Design · hard · Evidence 41/100How would you debug unexpectedly low GPU utilization during inference?ML System Design · hard · Evidence 41/100How would you engineer features from sparse professional profiles?ML System Design · medium · Evidence 41/100Cold Start: New Users vs New ItemsML System Design · medium · Evidence 41/100How would you separate offline and online pipelines in a large recommender?ML System Design · hard · Evidence 41/100How would you construct labels for a multimodal document prediction task?ML System Design · medium · Evidence 41/100How would you handle a missing image or missing text modality at inference time?ML System Design · hard · Evidence 41/100How would you design fallback behavior when a Q&A bot cannot retrieve enough evidence?GenAI & LLM · hard · Evidence 41/100How would you evaluate retrieval quality separately from answer quality in a Q&A bot?GenAI & LLM · hard · Evidence 41/100For a Q&A bot, when would you add RAG instead of relying on a parametric model alone?GenAI & LLM · medium · Evidence 41/100How do continuous batching and paged attention improve LLM serving?GenAI & LLM · hard · Evidence 41/100Explain the architecture of an LLM serving stack built around vLLM.GenAI & LLM · hard · Evidence 41/100Design text-to-image search with embeddingsGenAI & LLM · medium · Evidence 40/100Design concurrent LLM servingGenAI & LLM · hard · Evidence 37/100How would you use a two-tower model for candidate retrieval?ML System Design · hard · Evidence 37/100How would you evaluate a video recommender offline and online?ML System Design · hard · Evidence 37/100How would you choose features for a short-video recommendation system?ML System Design · medium · Evidence 37/100Design an Image Classifier for 500 Tree SpeciesML System Design · hard · Evidence 34/100Design a Second-Pass Feed RankerML System Design · hard · Evidence 34/100Design Hashtag-to-Video RelevanceML System Design · hard · Evidence 34/100Transformer vs RNNGenAI & LLM · medium · Evidence 34/100Design RAG with strict source controlGenAI & LLM · hard · Evidence 31/100