practice a little every day, paws crossed ♡
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 2551–2580 of 3640 results
- OpenAIMediumMachine Learning EngineerDesign an Agent Harness and Evaluation System
Design a safe harness and evaluation platform for a tool-using AI agent. Address isolated execution, repeatable task sets, trajectory tracing, deterministic and rubric-based graders, repeated trials, privacy, regression gates, and reward-hacking risks.
ML System DesignOnsite - OpenAIMediumSoftware EngineerDesign an Asynchronous Text-to-Video Generation Service
This question tests the ability to design a high-latency, asynchronous job-processing pipeline with a focus on fault tolerance, job state management, and reliable resource handling.
System DesignOnsiteAsync ProcessingScalabilityJob Queues - OpenAIHardMachine Learning EngineerDesign an Extensible Simulation Engine
This problem tests object-oriented design patterns and modular architecture by requiring the implementation of a simulation engine that remains agnostic to the specific rules of a two-player turn-based game.
Software Engineering FundamentalsOnsiteSystem DesignObject-Oriented DesignDesign Patterns - OpenAIHardSoftware EngineerDesign an Online Chess Platform: Elo Matchmaking, WebSocket Moves, Game Clocks
Design an online chess platform: match players of similar Elo rating quickly, deliver moves in real time over WebSocket connections, and enforce fair game clocks despite network latency and disconnections. It tests matchmaking, stateful real-time architecture, server-authoritative timing and failover.
System DesignOnsite - OpenAIMediumSoftware EngineerDesign an Online Chess Service from Launch to Scale
This problem tests the ability to design a real-time, stateful gaming system, focusing on concurrency control, low-latency communication, and fault tolerance.
System DesignOnsiteReal-timeConcurrencyWebSockets - OpenAIHardSoftware EngineerDesign an Online Multiplayer Chess Service with Real-Time Moves
Design an online chess service where two remote players play live games, every move is validated and delivered to the opponent immediately, and results are recorded. It tests authoritative game state, real-time transport, fair clocks, move ordering, recovery of games in progress after failures, and scaling to many concurrent games and spectators.
System DesignOnsite - OpenAIHardSoftware EngineerDesign an SSH-Based Cloud Development Environment
This question tests your ability to design a distributed system that manages stateful, long-running compute resources while ensuring isolation, high availability, and secure remote access.
System DesignOnsiteDistributed SystemsCloud InfrastructureSecurity - OpenAIHardMachine Learning EngineerDesign an image/video near-duplicate detection system
This question evaluates competency in ML system design and large-scale multimedia retrieval, focusing on perceptual fingerprinting versus embedding strategies, scalable indexing and nearest-neighbor retrieval, and robustness to resizing, re-encoding, watermarks, minor edits, and adversarial manipulations.
ML System DesignOnsite - OpenAIHardData ScientistDesign and Defend a Free-Trial Conversion Analysis
Walk through how to plan, execute, and defend a free-trial conversion analysis end to end: cohort definition, data-quality checks, standards for a credible insight, and a recommendation deck that survives live pushback.
Analytics & ExperimentationOnsiteexperimentationmetricscase-study - OpenAIHardMachine Learning EngineerDesign and optimize a RAG system
An OpenAI machine learning engineer onsite ML system design question: design, optimize and evaluate a production Retrieval-Augmented Generation (RAG) system that answers questions over a private corpus with citations. The merged answer covers the full pipeline — ingestion and layout-aware parsing, structure-aware chunking, embeddings and versioning, hybrid BM25 plus vector retrieval with RRF fusion, cross-encoder reranking, grounded generation with forced citations, abstention and a fallback ladder, ACL enforcement during retrieval, and the insert/edit/delete freshness lifecycle. It then gives a stage-by-stage evaluation plan (chunk answerability, hit-rate@K, recall@K, nDCG, faithfulness, citation quality, end-to-end success), how to bootstrap labels with no gold set, and follow-ups on debugging a wrongly cited document, embedding-model upgrades, multi-hop versus list-all queries, near-duplicates, conflicting sources and validating an LLM judge.
ML System DesignOnsite - OpenAIMediumSoftware EngineerDesign coffee-shop card payments with tips and nightly batch settlement at scale
Design a card payment system for coffee shops where purchases are authorized at the counter, tips are added afterwards, and transactions settle in a nightly batch. The discussion centers on scaling nightly settlement, along with idempotent APIs, failure handling and reconciliation against processor reports.
System DesignOnsite - OpenAIMediumSoftware EngineerDesign power grid device monitoring over an unreliable public internet
Design a monitoring system for power grid field devices that report their status over an unreliable public internet. It tests store-and-forward reporting, idempotent ingestion of late and duplicate data, telling silent devices from failed ones, alerting, and security for critical infrastructure.
System DesignOnsite - OpenAIMediumSoftware EngineerDesign webhook, POI, chat, CI/CD, payments
This question evaluates proficiency in distributed system design, covering scalability, availability, durability, data modeling, geospatial indexing, event-driven delivery, API design, fault tolerance, security, and operational monitoring for high-throughput backend services.
System DesignOnsite - OpenAIMediumMachine Learning EngineerExplain KV cache in Transformer inference
This question evaluates understanding of KV cache mechanisms in Transformer inference, including attention-state caching, memory and latency trade-offs, and engineering optimizations for autoregressive decoding.
Software Engineering FundamentalsOnsite - OpenAIHardMachine Learning EngineerExplain motivation and mission alignment
This question evaluates motivation, mission alignment, cultural fit, and communication skills by probing personal values, long-term goals, and the ability to articulate congruence with an organization's purpose.
Behavioral & LeadershipOnsite - OpenAIHardMachine Learning EngineerExplain what torch.distributed.barrier does
This question evaluates understanding of PyTorch distributed synchronization primitives, process coordination, and concurrency control in distributed training workflows.
Software Engineering FundamentalsOnsite - OpenAIHardSoftware EngineerExplain your perspective on AI safety
An OpenAI software engineer onsite behavioral question asking you to define AI safety in practical product terms, triage near-term versus long-term risks, and assess AI's effect on human work, the economy, and society. The answer must then connect those views to concrete engineering practice — threat modeling, guardrails, evals and red-teaming, phased rollout, and incident response. It also tests whether you can balance responsible deployment against shipping speed in both directions.
Behavioral & LeadershipOnsite - OpenAIMediumApplied ScientistExtensible card-hand ranking: three of a kind, flush, pair and a discard rule
A three-part object-oriented coding exercise built around a card game: rank hands by three of a kind, pair and high card, insert a flush category, then add a discard rule that removes hands that are not a pair, returning a log of each game. It tests extensible design, deterministic tie-breaking and absorbing rule changes cleanly.
Software Engineering FundamentalsOnsite - OpenAIHardSoftware EngineerFault-Tolerant Work Queue with Leases, Fencing Tokens and DAG Task Dependencies
Implement a fault-tolerant work queue in which workers claim tasks under expiring leases, crashed workers' tasks are reassigned, and late reports from stale workers are rejected, then add DAG dependencies with cycle detection. It tests leases, fencing tokens, topological release and delivery semantics.
Software Engineering FundamentalsOnsite - OpenAIMediumMachine Learning EngineerFilter Bad Human Annotations
Tests the ability to detect and filter low-quality human labels in large-scale dataset pipelines by modeling annotator reliability, distinguishing genuine ambiguity from noise, and evaluating downstream model performance.
Machine LearningOnsitedata-centric-airlhflatent-variable-models - OpenAIMediumMachine Learning EngineerFind Minimum Compatible Version
Implement an algorithm to find the earliest compatible software version while minimizing expensive compatibility-check API calls across partitioned version branches.
Coding & AlgorithmsOnsiteBinary SearchAlgorithmsOptimization - OpenAIHardSoftware EngineerFind a Hidden Number When Each Comparison Answer Arrives One Call Late
Find a hidden integer from 1 to n using a comparison API whose answers arrive one call late: each call submits a new guess but returns the result for the previous one. Build a correct search first, then reduce the worst-case number of calls. It tests binary search under delayed feedback and worst-case analysis.
Software Engineering FundamentalsOnsite - OpenAIMediumMachine Learning EngineerFind the First Working Version
This question tests the ability to combine custom comparison logic with a binary search approach to identify the first transition point in a monotonic sequence.
Coding & AlgorithmsOnsiteBinary SearchAlgorithmsOptimization - OpenAIHardSoftware EngineerFollow Graph with Cheap Snapshots and Top-K Follow Recommendations
Build a one-directional follow graph that supports cheap snapshots and queries against past snapshots, then recommend the top k accounts a user should follow. It tests versioned data structures, snapshot semantics, two-hop graph traversal, heap-based top-k selection and scaling to highly connected accounts.
Software Engineering FundamentalsOnsite - OpenAIHardSoftware EngineerGPU Credit Ledger with Expiring Grants, Out-of-Order Operations and Fast Balances
Implement a GPU credit ledger with expiring grants, subtractions that spend the soonest-expiring credits first, and balance queries, where operations can arrive out of timestamp order. Then avoid recomputing from time zero on every query. It tests event replay, priority queues and checkpoint invalidation.
Software Engineering FundamentalsOnsite - OpenAIMediumSoftware EngineerGPU Credit Ledger: Spend Earliest-Expiring Credits First, Stop on Overdraft
A coding and design question about an in-memory GPU credit ledger that grants expiring credits, charges usage against the credit that expires soonest, and returns no balance once any charge overdraws the account. It tests data structure choice, expiry and boundary conventions, and optimizing the ledger for a very large number of credit changes.
Software Engineering FundamentalsOnsite - OpenAIMediumSoftware EngineerGame Telemetry Windows: Match Counts, Live Sessions, Late Events
Build an in-memory service that counts per-match actions inside a global 10-minute window, tracks each player's live client sessions from heartbeats and logouts, and stays correct when events arrive late. Tests bounded-memory eviction, amortized analysis, lazy deletion and exact window semantics.
Software Engineering FundamentalsOnsitesliding-windowheapqueue - OpenAIHardMachine Learning EngineerHand-Rolled Backprop: Running Product and Stable Cross-Entropy
Derive and implement, without an autodiff library, the backward pass of a running (cumulative) product in linear time without division, then a numerically stable softmax cross-entropy and its gradient. Tests chain-rule fluency, prefix/suffix tricks, and gradient checking.
Machine LearningOnsiteautodiffbackpropagationnumerical-stability - OpenAIHardMachine Learning EngineerHand-Written Gradients: Prefix-Product Backward Pass and Stable Cross-Entropy
Write gradients by hand: the forward and backward pass of the prefix-product operation, in linear time and without division so inputs may contain zeros, and a numerically stable softmax cross-entropy loss with its gradient in NumPy. It tests the chain rule, prefix and suffix products, and gradient checking.
Machine LearningOnsite - OpenAIHardSoftware EngineerImplement Disease and Friend Snapshot Models
This question evaluates skills in graph-based modeling and temporal state simulation as well as the design and implementation of snapshot-capable directed graph data structures, covering competencies in propagation and state transitions (including immunity, recovery, and death), efficient snapshotting, querying, ranking, and diffing of follow relationships. It is commonly asked to assess algorithmic reasoning, correctness under edge cases, and complexity-aware implementation; the problem belongs to the Coding & Algorithms domain and tests both conceptual understanding of dynamic graph behavior and practical application-level implementation of data structures and immutable snapshots.
Coding & AlgorithmsOnsite
OpenAI