OpenAI interview questions

Showing 31–60 of 342 results

• Count Events in a Trailing 15-Minute Window with a Fixed-Size Ring BufferHard

  Implement an event counter that returns how many events occurred in the trailing 15 minutes, using a fixed-size ring buffer of per-second buckets so memory stays bounded at any event rate. It tests circular buffers, stale-slot detection, window boundaries, late events and thread safety.

  OpenAISoftware EngineerSoftware Engineering FundamentalsOnsite
• Count Machines and Recover a Distributed Tree TopologyMedium

  Implement a distributed tree discovery and aggregation protocol that counts reachable machines, recovers edges, handles network retries, detects unreachable subtrees, and validates tree invariants.

  OpenAISoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Count Ordered Four-Shop Road Paths That Visit One Shop of Each TypeHard

  Count the ordered routes through a town's road network that visit exactly one shop of each of four types, where consecutive shops must share a direct road and a route and its reverse count separately. Tests graph modeling, path enumeration under type constraints and careful counting on small graphs.

  OpenAISoftware EngineerCoding & AlgorithmsOnline Assessment
• Debug MiniGPT and Backpropagate MatmulMedium

  This question evaluates proficiency in debugging PyTorch transformer implementations, reasoning about attention mechanics and causal masking, implementing key–value caching for autoregressive decoding, and constructing provably correct backward passes for matrix multiplications and custom autograd.Functions.

  OpenAIMachine Learning EngineerMachine LearningTechnical Screen
• Debug Transformer and Add KV CacheMedium

  This problem tests your understanding of transformer architecture by requiring you to fix a broken attention mechanism and implement a Key-Value cache to optimize autoregressive inference.

  OpenAIMachine Learning EngineerMachine LearningOnsite
• Debug a Concurrent Job SchedulerMedium

  This question evaluates debugging and hardening skills for concurrent systems, focusing on detection and reasoning about race conditions, deadlocks, lock-contention hotspots, state-machine invariant enforcement, rate-limiting, retry semantics, and instrumentation for per-job metrics in a job scheduler.

  OpenAIMachine Learning EngineerSoftware Engineering FundamentalsTechnical Screen
• Debug a Machine Learning PipelineMedium

  This question evaluates a candidate's ability to diagnose production machine learning failures, covering competencies in data quality, data and concept drift detection, model versioning and deployment checks, and operational debugging within an MLOps context.

  OpenAISoftware EngineerMachine LearningTechnical Screen
• Debug a Small PyTorch Transformer with Four Bugs, Then Add a KV CacheHard

  Debug a small PyTorch transformer language model that hides four bugs, some of which crash and some of which silently ruin training, then add a key-value cache so generation computes only the newest token each step. It tests causal masking, positional encoding, training-loop order and cached decoding.

  OpenAIMachine Learning EngineerMachine LearningOnsite
• Debug a Transformer Implementation and Implement a KV Cache for DecodingMedium

  Practice ML coding question on Transformer language models: find and fix several deliberately planted bugs in a decoder-only Transformer implementation, then implement key-value caching from scratch for autoregressive generation. Tests attention mechanics, tensor-shape reasoning, causal masking, position handling, and verification against full recomputation.

  OpenAIMachine Learning EngineerMachine LearningTechnical Screen
• Debug a broken Transformer implementationHard

  This question evaluates a candidate's ability to debug and validate a Transformer implementation, focusing on attention masking, parameter initialization, loss alignment, and other implementation-level correctness issues.

  OpenAIMachine Learning EngineerMachine LearningOnsite
• Debug a failing ML classifierHard

  This question evaluates a candidate's ability to diagnose a poorly generalizing ML classifier, testing practical knowledge of data leakage, feature drift, and model calibration. It assesses applied machine learning debugging skills — including cross-validation design and error analysis — commonly tested in senior engineering and ML roles.

  OpenAISoftware EngineerMachine LearningTechnical Screen
• Debug a transformer training pipelineHard

  An OpenAI ML engineer technical-screen debugging question: given a PyTorch Transformer training pipeline whose loss diverges and validation accuracy stays near chance (plus mask IndexErrors, AMP crashes, and nondeterminism), identify the distinct bugs and fix them. It tests Transformer internals - attention/causal masking, teacher-forcing label shift, positional encodings, ignore_index loss, optimizer and GradScaler order, determinism, and DDP - along with writing unit tests and a structured debugging plan.

  OpenAIMachine Learning EngineerMachine LearningTechnical Screen
• Debug and harden trial-assignment Python codeMedium

  This question evaluates understanding of experiment assignment, data integrity, reliability engineering, concurrency, observability, and privacy in data systems, testing competencies around randomness, idempotency, time semantics, retries, and failure-mode identification.

  OpenAIData ScientistData Manipulation (SQL/Python)Technical Screen
• Debug transformer and train classifierHard

  This question evaluates debugging and practical implementation skills for transformer-based text classification, covering model correctness, training loop integrity, data preprocessing, and evaluation metrics within the Machine Learning / Natural Language Processing domain.

  OpenAIMachine Learning EngineerMachine LearningTechnical Screen
• Defend a Research Direction and Experiment DesignMedium

  This question evaluates a candidate's ability to synthesize the state of the art in Machine Learning, defend a research direction, and design rigorous experiments, measuring competencies in literature analysis, methodological justification, experimental design, and technical communication.

  OpenAIMachine Learning EngineerMachine LearningOnsite
• Derive Backpropagation for Matrix-Product LayersHard

  This question evaluates understanding of backpropagation through linear blocks, matrix calculus, and the computation of gradients with respect to individual weight matrices in a product of trainable matrices, including reasoning about tensor shapes and matrix differentials.

  OpenAIMachine Learning EngineerMachine LearningTechnical Screen
• Derive MLE and Bayesian posterior for BernoulliMedium

  This question evaluates parameter estimation and inferential reasoning for Bernoulli/binomial data, including maximum likelihood estimation and its asymptotic variance, Bayesian updating with a Beta prior and the resulting posterior and posterior predictive probability, as well as interval estimation; it is classified in the Statistics & Math domain. It is commonly asked to assess understanding of frequentist versus Bayesian inference, the behavior and reliability of asymptotic approximations and credible intervals, and the practical implications for predictive probability and uncertainty quantification, testing both conceptual understanding and practical application.

  OpenAIMachine Learning EngineerStatistics & MathOnsite
• Derive Sharded Matrix Multiplication and BackpropagationMedium

  Derive forward and backward computation for column-sharded matrix multiplication across multiple devices. Track every local tensor shape, identify the collective needed for replicated input gradients, and propose tests for shard ordering, scaling, and numerical errors.

  OpenAIMachine Learning EngineerMachine LearningOnsite
• Describe handling pressure and present your workMedium

  Describe handling pressure and present your work evaluates behavioral evidence, ownership, communication, trade-offs, and measurable outcomes in a realistic interview setting. A strong answer states assumptions, handles edge cases, explains trade-offs, and shows how to validate the result clearly.

  OpenAIMachine Learning EngineerBehavioral & LeadershipTechnical Screen
• Describe relevant PM experienceMedium

  Prepare OpenAI PM behavioral answers on analytical tools, mobile feature optimization, localization, and experiments. Includes STAR-style examples, metrics, guardrails, and tailoring advice.

  OpenAIProduct ManagerBehavioral & LeadershipHR Screen
• Design Android MVVM API ArchitectureMedium

  Design a robust, scalable Android client architecture using MVVM and Unidirectional Data Flow (UDF) to interact with remote REST and streaming APIs, handling configuration changes, lifecycle events, and coroutine cancellation.

  OpenAIAndroid EngineerSystem DesignTechnical Screen
• Design Bakery Chain Card Payments With Tips and Nightly SettlementMedium

  Design the payment system for a bakery chain's counter terminals, where a card is authorized for the purchase, a tip is added afterward, and each location's transactions are captured together in a nightly settlement batch that must scale as the chain grows.

  OpenAISoftware EngineerSystem DesignOnsite
• Design CI/CD Build CachingMedium

  This problem evaluates your ability to design a distributed caching system for CI/CD pipelines, focusing on cache key strategies, invalidation, and storage scalability.

  OpenAISoftware EngineerSystem DesignTechnical Screen
• Design Command Dispatch and Telemetry Reconciliation for Unreliable DevicesHard

  Design a resilient control plane and telemetry reconciliation system for millions of intermittently connected edge devices, ensuring safe command execution, idempotency, and accurate state tracking despite network unreliability.

  OpenAISoftware EngineerSystem DesignTechnical Screen
• Design Duplicate File DetectionMedium

  This question tests the ability to design a scalable content-based deduplication system using cryptographic hashing and distributed storage principles to manage large-scale file ingestion.

  OpenAIMachine Learning EngineerSystem DesignOnsite
• Design First-Fit and Best-Fit Memory AllocationMedium

  Design an in-memory heap allocator supporting first-fit and best-fit allocation, block splitting, adjacent free block coalescing, and safe handle validation to prevent stale or forged free operations.

  OpenAISoftware EngineerSoftware Engineering FundamentalsTechnical Screen
• Design GPU Scheduling for a Video Generation PlatformMedium

  This question tests your ability to design a distributed job scheduling system that manages heterogeneous GPU resources for computationally expensive, variable-length video generation tasks.

  OpenAISoftware EngineerSystem DesignOnsite
• Design GPU credit allocatorHard

  A multi-tenant GPU credit allocation system design (OpenAI software-engineer system-design screen): track and deduct GPU credits in real time across many nodes, issue/transfer/spend credits via idempotent APIs, and enforce budgets, hierarchical quotas, rate limits, and fair-share with scheduler integration. Covers prepaid vs postpaid, an immutable double-entry ledger with reservations, overspend prevention, failure recovery, scaling, and observability.

  OpenAISoftware EngineerSystem DesignTechnical Screen
• Design IDE Sandbox and PaymentsHard

  This question evaluates system design competencies such as distributed systems architecture, scalability, isolation and security, API and state-machine design, storage and synchronization, resiliency, observability, and compliance applied to two domains: a cloud IDE sandbox and a payment processing backend, with domain: System Design.

  OpenAISoftware EngineerSystem DesignOnsite
• Design LLM search handling long token inputsHard

  Design an LLM search and RAG system that handles long documents and long user queries. Covers chunking, hierarchical retrieval, query compression, context budgeting, citations, latency, and evaluation.

  OpenAIMachine Learning EngineerML System DesignOnsite