practice a little every day, paws crossed ♡
Machine Learning Engineer Interview Questions
Practice 9403 coding, system design and machine learning questions from 546 companies, sorted by role, topic and interview round. Every question has step-by-step hints, a code runner that supports 8 languages and an AI mentor; members also get the full solution walkthrough.
- 9403
- Questions
- 546
- Companies
- 11
- Topics
- 8
- Languages
Showing 631–660 of 782 results
- SnapchatHardMachine Learning EngineerExplain Core ML Training, Evaluation, and Transformer Concepts
Tests end-to-end understanding of machine learning fundamentals, covering regularization against overfitting, handling severe class imbalance, optimizer mechanics, threshold calibration, and Transformer self-attention dynamics.
Machine LearningTechnical Screenmachine-learningdeep-learningtransformers - SnapchatHardMachine Learning EngineerExplain Embeddings, Positive Labels, and Retraining in a Recommendation Project
Discuss recommendation-project ownership, embeddings, positive labels and retraining with attention to exposure bias and data leakage.
Machine LearningTechnical Screen - SnapchatHardMachine Learning EngineerExplain LLM tuning and transformer basics
This question evaluates competencies in large language model tuning and Transformer fundamentals, covering fine-tuning strategies, dataset construction and labeling, model adaptation choices, loss functions and evaluation metrics, regularization techniques, optimizer selection, self-attention and multi-head attention, and the end-to-end Transformer decoder mathematical steps. It is commonly asked in Machine Learning interviews because it probes both conceptual understanding and practical application—examining architectural trade-offs, optimization and regularization decisions, deployment constraints, and reasoning about failure modes such as overfitting and hallucination within the Machine Learning domain.
Machine LearningTechnical Screen - SnapchatHardMachine Learning EngineerExplain Neural-Network Regularization and Dropout
Explain L1, L2, dropout scaling, train/inference behavior, early stopping, and the difference between gradient penalties and decoupled weight decay.
Machine LearningOnsite - SnapchatMediumMachine Learning EngineerExplain Overfitting and Transformer Attention
This question evaluates understanding of model generalization and regularization techniques alongside Transformer self-attention and positional encoding, assessing competencies in diagnosing overfitting, applying appropriate mitigation strategies, and interpreting attention mechanisms.
Machine LearningTechnical Screen - SnapchatHardMachine Learning EngineerExplain Vanishing Gradients in Neural-Network Training
Diagnose vanishing gradients through the chain rule, activation and gradient checks, and architecture-appropriate training remedies.
Machine LearningTechnical Screen - SnapchatMediumMachine Learning EngineerFind K Nearest Points
This question tests the ability to efficiently find the k-closest points to an origin in a 2D plane, focusing on heap-based or selection-based data structures.
Coding & AlgorithmsOnsiteHeapGeometryQuickSelect - SnapchatMediumMachine Learning EngineerFind Shortest Clear Grid Path
Solve the shortest clear path in an n by n binary grid using BFS in 8 directions. Covers blocked endpoints, visited marking, path length, correctness, complexity, and weighted-move follow-ups.
Coding & AlgorithmsOnsite - SnapchatMediumMachine Learning EngineerFind all quadruplets summing to target
This problem tests your ability to optimize nested loops using a two-pointer approach to achieve a time complexity better than brute force when searching for subsets of a fixed size that sum to a target value.
Coding & AlgorithmsOnsiteArraysTwo PointersSorting - SnapchatHardMachine Learning EngineerFind the Smallest Range Covering K Sorted Lists
Tests multi-pointer tracking and priority queue fundamentals to find the narrowest bounding interval containing at least one element from each of k sorted lists.
Coding & AlgorithmsTechnical ScreenHeapSliding WindowGreedy - SnapchatHardMachine Learning EngineerImplement calculator and graph connectivity
This question evaluates proficiency in parsing and arithmetic expression evaluation (including operator precedence and nested parentheses) alongside implementation of disjoint-set (Union-Find) data structures for computing graph connected components.
Coding & AlgorithmsTechnical Screen - SnapchatHardMachine Learning EngineerRender an Organization Chart and Find Skip-Level Pairs
Build and render an organization tree with ordered indentation, infer its root, and return exactly two-level manager-descendant pairs.
Coding & AlgorithmsOnsite - SnapchatHardMachine Learning EngineerWhat Happens When All Neural-Network Weights Start at Zero?
Explain zero-weight initialization through hidden-unit symmetry and gradient flow, and contrast it with linear and logistic models.
Machine LearningTechnical Screen - SnowflakeMediumMachine Learning EngineerCore Components of LLM Agent Memory: What Each Stores and How It Is Written and Read
Explain the common components of an LLM agent's memory, from the context window to session, episodic, semantic and procedural memory, and how each is written, retrieved and maintained. It tests agent architecture knowledge, context budgeting, conflict handling, forgetting and memory security.
Machine LearningOnsite - SnowflakeMediumMachine Learning EngineerDesign a Pipeline That Distills Reusable Agent Skills for Enterprise Customers
Design a pipeline that automatically extracts reusable agent skills for enterprise customers from traces and documents, validates them and delivers them to agents. It tests LLM system design, outcome labeling, offline evaluation against a baseline, tenant isolation, versioning and drift monitoring.
ML System DesignOnsite - SnowflakeMediumMachine Learning EngineerDesign and Implement a Rate Limiter with an AI Assistant, Then Defend and Test It
Design and implement a rate limiter in an AI-assisted coding round: choose and defend an algorithm against its alternatives, direct an AI assistant to write the code, review it and test it. It tests rate-limiting algorithm trade-offs, concurrency, time handling and critical review of generated code.
Software Engineering FundamentalsOnsite - SpacexMediumMachine Learning EngineerDiagnose a Production Classifier and Defend Its Systems Design
This problem tests the ability to diagnose training-serving skew in machine learning models and the architectural considerations required to maintain consistency between offline evaluation and online production environments.
Machine LearningTechnical ScreenMLOpsSystem DesignTraining-Serving Skew - SpotifyMediumMachine Learning EngineerDesign Podcast Recap Generation
This question evaluates system-design and machine-learning engineering competencies, including streaming versus batch ingestion, audio transcription and chunking, long-context retrieval and prompt/fine-tuning choices, model serving and cost-latency trade-offs, storage and indexing of transcripts and embeddings, evaluation of factual accuracy, and operational monitoring and recovery. Commonly asked to assess the ability to balance latency, throughput, cost, and accuracy in production ML pipelines, it is categorized as ML System Design and tests both conceptual understanding of trade-offs and practical application of scalable, reliable architecture.
ML System DesignOnsite - SpotifyMediumMachine Learning EngineerDesign a Sequential Personalized Playlist Recommender
This problem tests your ability to design a production-grade recommendation pipeline that balances immediate user intent (sequential context) with long-term personalization, while managing latency constraints at scale.
ML System DesignOnsiteRecSysML-ArchitectureLatency - SpotifyMediumMachine Learning EngineerJoin Tables and Rank Users
This question evaluates proficiency in SQL and data-engineering skills—specifically joins, aggregation and ranking via window functions, event deduplication, and query performance considerations across analytics tables.
Coding & AlgorithmsOnsite - SpotifyHardMachine Learning EngineerPredict Next-Week Streaming Engagement at Event Scale
This question tests the ability to design an end-to-end ML system for user retention prediction, focusing on feature engineering, model scalability for massive user bases, and real-time inference latency.
ML System DesignOnsiteSystem DesignMLOpsRecommendation Systems - SpotifyEasyMachine Learning EngineerQuery Top Played Tracks Globally And By Country
Tests writing clean SQL queries with aggregations, window functions, and deterministic tie-breaking to extract top playback metrics globally and per country.
Data Manipulation (SQL/Python)Technical Screensqlwindow-functionsaggregation - StackAdaptMediumMachine Learning EngineerCamel and Bananas: Maximize Bananas Delivered Across the Desert
Classic camel-and-bananas brainteaser: a camel that carries at most 1,000 bananas and eats one per kilometer must move 3,000 bananas to a market 1,000 km away. Find the maximum number that can arrive, testing piecewise cost reasoning, depot planning, and a proof of optimality.
Statistics & MathOnsite - StackAdaptEasyMachine Learning EngineerExplain Why a Geometric Mean Can Exceed Its Median
Tests derivation and comparison of the expected value and median of a geometric distribution, explaining how right-skewness drives the mean above the median across different parameterizations.
Statistics & MathTechnical Screenprobabilitystatisticsgeometric-distribution - StackAdaptMediumMachine Learning EngineerFind the Flaws in an End-to-End ML Pipeline Case Study Report
Review a condensed end-to-end machine learning case study, from its goal and labels through normalization, evaluation, and deployment, and point out what is wrong with it. Tests systematic pipeline review, reasoning about what data each step uses, and ranking problems by how much they undermine the conclusions.
Machine LearningOnsite - StackAdaptEasyMachine Learning EngineerMaintain the Median of an Integer Stream
This problem tests your ability to maintain dynamic statistics in a data stream using efficient heap-based data structures.
Coding & AlgorithmsTechnical ScreenHeapsData StreamsAlgorithms - StackAdaptMediumMachine Learning EngineerRebuild a Binary Tree From Preorder and Inorder Sequences, Returned in Level Order
Reconstruct a binary tree with distinct values from its preorder and inorder traversals and return it in level-order form with null markers. Tests how the two traversal orders pin down structure, careful index bookkeeping, and efficient handling of deep, skewed trees.
Coding & AlgorithmsOnsite - StackAdaptMediumMachine Learning EngineerReservoir Sampling: Uniform k-of-n Sample From a Stream of Unknown Length
Design a one-pass algorithm that keeps a uniform random sample of k items from a stream whose length is unknown in advance, using memory proportional to k. Tests the replacement rule, an inductive proof that every item is kept with equal probability, complexity, and edge cases.
Statistics & MathOnsite - Startups.ComMediumMachine Learning EngineerDebug and scale a PyTorch training loop
This question evaluates debugging and performance engineering skills for deep learning training pipelines, including diagnosing convergence and instability issues in PyTorch, scaling training with Fully Sharded Data Parallel (FSDP), designing sparse gradient all-reduce communication patterns, and judging when to implement custom CUDA/Triton kernels. Commonly asked for Machine Learning Engineer roles within the Software Engineering Fundamentals domain, it assesses both conceptual understanding of numerical stability and distributed systems and practical application skills in configuring multi-GPU training, communication trade-offs, and low-level performance and correctness validation.
Software Engineering FundamentalsOnsite - Startups.ComMediumMachine Learning EngineerDesign efficient Transformer inference with KV cache
This question evaluates an engineer's competency in ML systems engineering, specifically understanding of decoder-only Transformer inference, KV cache semantics and which tensors are cached per layer, attention mechanics, memory layout, and correctness under branching and long-context workloads.
ML System DesignOnsite
Snapchat
Snowflake
Spotify
StackAdapt