Search the siteSearch in

Interview paths
Question bank
Coding practice
Study materials
Infrastructure lab

Community

Weekly sessions
Peer practice
Contribute

News

Jobs

Open roles
Market intelligence
Companies
Roles

Feedback

Product feedback
Developer feedback

Checking account…

Home
Question bank

Frontier AI technical interview question bank.

Practice coding, system design, behavioral, and project questions for frontier AI roles.

Company

Loading the question catalog…

Preparing this page…

Your current navigation and account state stay available while the next page loads.

Frontier Rehearsal · Independent practice · Not affiliated with OpenAI, Anthropic, or xAI.

Home Interview questions Company interview processes Company reports Role reports Market intelligence Open roles Python lab Materials How it works About Feedback Uploads Feature requests Terms Privacy 中文 English

Home
Question bank

Frontier AI technical interview question bank.

Practice coding, system design, behavioral, and project questions for frontier AI roles.

Company

FREE PRACTICE

Public demos, always pinned.

Start with either demo. They remain here when you filter or sort.

Free account: sign in to practice the marked Free questions.

Newest published questions; deterministic, never randomized.

Demo 1Number search with delayed feedback
Demo 2Design real-time device monitoring and control

Question bank filters

Reset filters

Question groups

Filter question bank by interview groupAll 269 Coding 162 System design 48 Behavioral 29 Hiring manager 16 Project deep dive 14

Topics

Filter question bank by topic tagsAll 269 Python 161 Algorithms 40 Data structures 46 Concurrency 27 Distributed systems 7 Storage & state 43 Networking 7 Reliability 30 ML / AI 27 Data / SQL 24 Security 7 Product design 25 Communication 66

Independent and not affiliated with the companies represented. Collected material does not predict future interview questions.

255 questions

Sort question bankSortFree first Newest Oldest AI ratio ↓AI ratio ↑

Question bank results

Every question has a public brief with the task, constraints, and follow-ups. Labels describe interactive practice access.

Public brief

Topic tags
Python
Storage & state
Reliability

OpenAICoding

Python codingWebAssembly + public tests2 versions

Reported interview periodNov 2025–Aug 2026

Persist a string map across restart

Implement a durable string-to-string key-value store using the provided FileSystem abstraction.

Practice focusIn the base task, shutdown() commits a complete snapshot using FileSystem.save blob(bytes) . A newly constructed store starts empty; restore() recovers the latest saved snapshot. The supplied storage replaces a whole blob atomically and remains alive across store instances. It models snapshot persistence, not partial writes or a real disk. Keys and values may contain arbitrary Unicode characters, including newlines; use a length-prefixed binary format rather than delimiter-only encoding. Do not use JSON or pickle.…

Public brief

Topic tags
Python
Storage & state

OpenAICoding

Python codingWebAssembly + public tests

Reported interview periodDec 2025–Aug 2026

Turn-Based Creature Battle

Implement a deterministic battle between two teams of creatures. Team A acts first. On each turn, the first living creature on the active team attacks the first living creature on the opposing team.

Practice focusRequirements - Damage reduces health; a creature with health <= 0 is knocked out and cannot act. - Alternate teams after every attack, skipping a side only if it has no living creature—at which point the battle ends. - Log every attack, knockout, and final winner in a deterministic format agreed before coding. - Do not mutate caller-owned teams unless the API explicitly documents it. - Validate empty teams, non-positive attack values, and battles that cannot make progress. Follow-ups 1. Order turns by speed rather than alternating teams. 2.…

New public brief

Topic tags
Python
Data structures

AnthropicCoding

Python codingWebAssembly + public tests

Reported interview periodJul 2025–Aug 2026

Reconstruct Stack Transitions from Samples and Events

Build utilities for converting between stack observations and call events. Stack frames are ordered outermost to innermost.

Practice focusFor an event stream, push on enter / start , pop on exit / end , record the stack after each event, and run-length encode adjacent identical stacks. For sampled stacks, compare adjacent samples using their longest common prefix. Emit removed frames as end events from innermost outward, then emit added frames as start events from outermost inward. Use the new sample's timestamp and leave frames in the final sample open. Samples have ts and stack fields; events emitted from samples have name , kind ( start or end ) and ts .…

Public brief

Topic tags
Python

MetaCoding

Python codingWebAssembly + public tests

Reported interview periodJan 2026–Aug 2026

Leftmost local maximum in a numeric sequence

Given a sequence whose adjacent values never tie, locate the first element that rises above every neighbor it has. Report that element's position.

Practice focusRequirements - The input is one line of space-separated integers with no count prefix, and it holds at least one value. - A position is a local maximum when its value is strictly greater than each neighbor that exists; the first and last positions have only one neighbor. - Return the zero-based index of the earliest such position as a decimal string. - A single-element sequence qualifies at index 0 . - Because neighbors never tie, at least one local maximum always exists and no equality rule is needed.…

Public brief

Topic tags
Python

MetaCoding

Python codingWebAssembly + public tests

Reported interview periodMar 2026–Aug 2026

Count contiguous runs with a target sum

Given an integer array and a target, count the non-empty contiguous runs whose values add up to that target. Values may be zero or negative, so a qualifying run can begin and end anywhere.

Practice focusRequirements - The first line holds the element count and the target, separated by a space; the second line holds the values. - A variant layout puts only the count on the first line, the values on the second, and the target on the third; read the target from whichever position it occupies. - Count every start and end pair, including single elements and runs whose values are all zero. - The empty run never counts, and overlapping runs are counted separately. - Return the count as a decimal string with no other text.…

Public brief

Topic tags
Python
Data structures
Storage & state
Networking

OpenAICoding

Python codingWebAssembly + public tests

Reported interview periodJan 2026–Aug 2026

Follow graph with snapshots

Implement a directed social graph with add_user, follow, and create_snapshot.

Practice focusA snapshot must preserve the follow relationships that existed when it was created. Later mutations to the live graph must not change earlier snapshots. A snapshot supports is following , get following , get followers , and recommend . Self-follows and repeated follows are no-ops. Rank recommendations by the number of two-hop mutual connections in descending order, then by user id in ascending order; return at most k users. Exclude the requesting user and accounts they already follow. Follow-ups 1. Add unfollow while preserving earlier snapshots. 2.…

Public brief

Topic tags
Python
Algorithms
Security

DatabricksCoding

Python codingWebAssembly + public tests

Reported interview periodApr 2026–Aug 2026

Delete a Position from Interval Coverage

A compact coverage set is represented by sorted, non-overlapping, non-touching inclusive intervals. If the covered values were expanded into one ordered sequence, delete the value at a given zero-based position…

Practice focusFor example, [(4, 7), (10, 11), (13, 15)] represents 4,5,6,7,10,11,13,14,15 . Deleting position 2 removes value 6 , producing [[4, 5], [7, 7], [10, 11], [13, 15]] . Requirements - Return intervals as [start, end] pairs. - Preserve sorted, non-overlapping, non-touching output. - A deletion may shrink an endpoint, remove a singleton interval, or split an interval in two. - Assume index refers to an existing covered value; state how you would report an invalid index. - Do not expand every covered integer when ranges may be large.…

Public brief

Topic tags
Python
Concurrency
Reliability

DatabricksCoding

Talk-through codingNo local grader

Reported interview periodNov 2025–Aug 2026

Durable Multi-Threaded Log Writer

Implement a log writer that accepts binary records from multiple threads and durably appends each complete record to a file.

Practice focusRequirements - Encode each record as a four-byte big-endian length followed by its payload. - push returns only after all bytes for that record are written and the file is synchronized to durable storage. - A record's bytes must never interleave with another record. - Calls from one thread retain program order; calls from different threads may appear in either order. - close must coordinate with active or future pushes and be idempotent. - Handle partial writes, I/O errors, and payload lengths that cannot be represented.…

Public brief

Topic tags
Python
Concurrency

DatabricksCoding

Python codingWebAssembly + public tests

Reported interview periodNov 2025–Aug 2026

Hybrid Run-Length and Bit-Packing Codec

Implement a reversible encoder for an integer sequence using run-length blocks for long runs and fixed-size literal blocks otherwise.

Practice focusRequirements and format - A run of at least eight equal values becomes ["RLE", value, count] and consumes the full run. - Otherwise consume the next eight values as ["BP", values] . - A final block may contain fewer than eight values. If every remaining value is equal, encode it as RLE; otherwise use a short BP block. - Walk strictly left to right, making output deterministic. decode must validate the entire block sequence before reconstructing it.…

Public brief

Topic tags
Algorithms
Data structures

OpenAISystem designAI-generated

Discussion / designResponse prompt

Reported interview periodJan 2026–Aug 2026

Crossword Constraint Solver

Given a crossword grid, a set of across/down clue entries, and candidate words for each clue, fill the grid so every entry uses one candidate and all crossing letters agree.

Practice focusRequirements - Define data structures for grid cells, clue slots, candidate domains, and crossing constraints. - Return one valid completed grid or report that none exists. - Use constraint propagation and backtracking rather than enumerating every combination. - Choose a search order such as minimum remaining values, and explain tie-breaking. - Restore domains correctly after a failed branch. - Handle prefilled letters, duplicate candidate words, disconnected regions, and ambiguous puzzles.…

New public brief

Topic tags
Python
Concurrency
ML / AI

AnthropicCodingHalf AI-generated

Python codingWebAssembly + public tests

Reported interview periodSep 2025–Aug 2026

Build a Tool-Using Research Agent

You are given an LLM client whose responses may contain text blocks or one or more tool-call blocks. Build an agent that answers a user question by repeatedly calling the model, executing requested tools, retur…

Practice focusAssume two tools are available: a stock-price lookup and a shares-outstanding lookup. A question may require one lookup, several independent lookups, or combining results such as market capitalization. Requirements - Preserve the complete conversation history, including assistant tool calls and matching tool results. - Execute every tool call in a response; do not silently discard parallel calls. - Match each result to its tool-call identifier. - Stop only on a final model response, and return its text.…

New public brief

Topic tags
Python
Data structures
Storage & state

AnthropicCoding

Talk-through codingNo local grader

Reported interview periodApr 2026–Aug 2026

Convert Sampled Call Stacks into Events

A profiler periodically records the active call stack. Convert these snapshots into the minimal sequence of call-start and call-end events.

Practice focusEach sample has an integer timestamp and a stack ordered from outermost to innermost. Each output event contains the sample timestamp, an event type ( start or end ), and a frame name. Rules - Compare each stack with the previous stack using the longest common prefix. - End removed frames from innermost to outermost. - Start added frames from outermost to innermost. - Adjacent identical stacks produce no events. - Equal names at different depths represent different invocations. - Do not invent end events for calls still active in the last sample.…

Public brief

Topic tags
Python
Networking
Security

DatabricksCoding

Python codingWebAssembly + public tests

Reported interview periodNov 2025–Aug 2026

Ordered IPv4 Firewall Rules

Implement an ordered firewall policy for IPv4 addresses.

Practice focusEach rule is [action, cidr] , where action is "ALLOW" or "DENY" . Evaluate rules from top to bottom; the first network containing the address determines the result. If no rule matches, deny access. A rule without a prefix length represents one address, equivalent to /32 . Requirements - Parse and validate dotted-decimal IPv4 addresses and CIDR prefix lengths. - Correctly handle /0 , /32 , network boundaries, and overlapping rules. - Preserve first-match semantics; a more specific later rule does not override an earlier match.…

Public brief

Topic tags
ML / AI
Product design
Communication

OpenAISystem designAI-generated

Discussion / designResponse prompt

Reported interview periodAug 2026

Design an interactive model-backed product feature

A collaborative writing product wants an assistant that can transform a selected passage according to a user instruction. The interaction should feel immediate, stream useful output, preserve user control, and …

Practice focusScenario A collaborative writing product wants an assistant that can transform a selected passage according to a user instruction. The interaction should feel immediate, stream useful output, preserve user control, and remain dependable when the model is slow, unavailable, or returns unsafe content. Design the feature from the client interaction through model serving and operational controls. Product requirements - Start visible progress quickly and stream partial output when possible.…

Public brief

Topic tags
Communication

OpenAIBehavioralAI-generated

Discussion / designResponse prompt

Reported interview periodDec 2025–Aug 2026

Recruiter screen: motivation, fit, and career narrative

Prepare for a concise recruiter conversation that establishes why you are exploring this role, whether the opportunity matches your goals, and whether practical constraints align. This is not a technical quiz; …

Practice focusScenario Prepare for a concise recruiter conversation that establishes why you are exploring this role, whether the opportunity matches your goals, and whether practical constraints align. This is not a technical quiz; the goal is a clear, credible conversation that helps both sides decide whether to continue. Prepare these responses 1. A 60–90 second summary of your background and the thread connecting your recent work. 2. Why this company and role are specifically interesting now. 3. What you want in your next position and what you do not want. 4.…

Public brief

OpenAISystem designAI-generated

Discussion / designResponse prompt

Reported interview periodAug 2026

Engineering Management Fundamentals

Describe how you operate an engineering team and support both delivery and people growth. Use specific examples rather than generic management principles.

Practice focusRequirements Prepare concise stories covering: - how you set goals, plan work, and run the team's weekly operating cadence; - how you hire and define a consistent technical bar; - how you coach strong performers and create growth opportunities; - how you diagnose and address sustained underperformance fairly; - how you balance technical quality, product urgency, and team health; - a decision you got wrong and what changed afterward.…

Public brief

Topic tags
Python
ML / AI
Product design

AnthropicCodingHalf AI-generated

Python codingWebAssembly + public tests

Reported interview periodSep 2025–Aug 2026

Implement Batched Scaled Dot-Product Attention

Implement attention for batched multi-head NumPy arrays on CPU. Inputs are finite arrays with the same floating-point dtype; dimensions are positive.

Practice focusShapes are q: (B, H, Tq, Dk) , k: (B, H, Tk, Dk) , and v: (B, H, Tk, Dv) . Return an array of shape (B, H, Tq, Dv) . When return weights=True , return (output, weights) , with weights of shape (B, H, Tq, Tk) . Requirements - Compute scores as q @ k.transpose(-2, -1) / sqrt(Dk) . - Apply softmax along the key-position dimension. - Support an optional boolean mask broadcastable to the score tensor, where false entries cannot receive probability. - Avoid numerical overflow by using a stable softmax or the framework's stable implementation.…

Public brief

Topic tags
Python
Data structures
Concurrency

OpenAICoding

Python codingWebAssembly + public tests2 versions

Reported interview periodDec 2025–Aug 2026

First-fit heap blocks

Implement a memory allocator over a fixed contiguous address range. Maintain free blocks in a doubly linked list ordered by starting address.

Practice focusallocate(size) must select the first block large enough for the request. Remove an exact-fit block; otherwise advance its start address and reduce its size. Return the allocated start address, or -1 when no block can satisfy the request. free(address, size) returns a region to the free list. Insert it in address order and coalesce it with an adjacent block on the left, right, or both sides when their boundaries touch. Preserve list ordering and links after every operation. Follow-ups 1. Reject overlapping or double frees without corrupting state. 2.…

Public brief

Topic tags
Python

AnthropicCoding

Python codingWebAssembly + public tests

Reported interview periodOct 2025–Aug 2026

Build and Scale an Image-Transformation Pipeline

Given a directory of source images and a directory of transformation specifications, produce one output for every image/specification pair. Each specification contains an ordered sequence selected from grayscal…

Practice focusUse the provided naming callback for every result. Apply transformations in order and save valid output images. Scaling truncates the computed dimensions to integers; rotation expands the canvas. Requirements and edge cases - Discover supported image and specification files recursively or according to the supplied listing helper. - Do not share mutable image objects across independent pipelines. - Ensure all scheduled work completes before returning and propagate worker failures.…

Public brief

Topic tags
Python
Data structures

AnthropicCoding

Talk-through codingNo local grader

Reported interview periodSep 2025–Aug 2026

Find Duplicate Files Efficiently

Walk a directory tree and return groups of files whose complete byte contents are identical.

Practice focusEach returned path must be absolute and resolved. Omit groups containing only one file. Group and path order are not significant. Requirements - Skip symbolic links and continue past files that become unreadable. - Treat multiple empty files as duplicates. - Avoid reading every byte of every file when cheap evidence can rule out equality. - A full-content comparison or collision-resistant digest must confirm every reported group. - Stream large files in bounded-size blocks rather than loading them entirely into memory.…

Public brief

Topic tags
Python
Data structures
Storage & state

OpenAICoding

Python codingWebAssembly + public tests

Reported interview periodNov 2025–Aug 2026

Iterator that can pause and resume

Implement ResumableListIterator, an iterator that can save and restore its progress.

Practice focusget state() must return a JSON-serializable value that fully identifies the next item to produce. set state(state) must restore that position so iteration continues without repeating or skipping values. The iterator should use constant additional memory and follow normal Python iterator behavior at end of input. Handle empty and single-item lists, restoration at the beginning and end, invalid state, and repeated save/restore cycles. Follow-ups 1. Extend the design to a nested iterator. 2. Detect restoration against a changed underlying collection. 3.…

Public brief

Topic tags
Python
Data / SQL

OpenAICoding

Python codingWebAssembly + public tests

Reported interview periodApr 2026–Aug 2026

Online softmax entropy

Implement numerically stable softmax, log-softmax, and entropy for a vector of logits. The standard entropy is -Σ p log p.

Practice focusThen implement a block-wise variant that processes the logits in bounded-size chunks without losing numerical stability. The block-wise result must match the full-vector computation within floating-point tolerance. Handle very large and very small logits, empty inputs according to the provided contract, and both supported entropy-sign conventions. Explain why subtracting the maximum logit prevents overflow and how partial statistics combine across blocks. Follow-ups 1. Return gradients with respect to the logits. 2.…

Public brief

OpenAISystem design

Discussion / designResponse prompt

Reported interview periodAug 2026

Token-level streaming differ with rollback

Design an incremental processor for a token stream whose producer may append new tokens or roll back previously emitted tokens.

Practice focusThe processor must always expose the latest sequence while doing work proportional to the changed suffix rather than rebuilding all derived state after every update. Define the update interface and the state needed to support append, rollback, and replacement. Explain how you would handle duplicate updates, rollback beyond retained history, bounded memory, persistence, and consumers that are temporarily offline. Follow-ups 1. Attach versions so consumers can reject stale or duplicate updates. 2.…

Public brief

Topic tags
Python
Algorithms
Concurrency
Distributed systems
ML / AI

OpenAICodingHalf AI-generated

Python codingWebAssembly + public tests

Reported interview periodJul 2026–Aug 2026

Debug sharded matrix-multiplication gradients

Implement forward and backward matrix multiplication across world workers.

Practice focusSplit A evenly by rows, replicate B , compute each local A i @ B , and concatenate the local outputs in row order. For the backward pass, compute each local dA i = dY i @ B.T , concatenate the dA i blocks, and compute the global dB as the sum of A i.T @ dY i across workers. Return results that match an unsharded NumPy implementation. Validate matrix shapes and explain the communication required by the forward and backward passes. Follow-ups 1. Shard B instead and compare communication costs. 2. Add uneven row partitions and preserve output ordering. 3.…

Previous7 of 11Next

Browse all original practice

Build a token-budget batcher
Sliding-window rate limiter
Interval calendar
Nested dictionary transforms
Command-log store
Dependency-ready scheduler
Streaming metric window
Evaluation slice aggregator
Build a reliable job runner
Build one adaptable decision story
Defend your role fit
Prepare one project deep dive
Design an evaluation artifact registry
Defend a reversible launch decision
Negotiate an ambiguous role scope
Trace a project from claim to evidence
Choose between two urgent priorities
Resolve a cross-functional disagreement
Own a consequential failure
Increase a team’s leverage
Warm a cold cache without a thundering herd
Ship a 500 GB checkpoint without a startup stampede
Make every document decision traceable to evidence
Share eight GPUs between two workloads with different promises
A mutable indexed sequence that must restart exactly
Trace a falling unit price to a rising bill
Keep a very long conversation usable while it streams
Design durable scheduled publishing
Make speculative code edits safe to accept
Recover a grouping rule from two-word phrases

CONTENT BOUNDARY

Independent frontier AI interview practice. Includes adaptations of collected interview material and original exercises; not official company questions and not affiliated with the companies represented. Collected material does not predict future interview questions. Report a correction.

Loading the question catalog…

FREE PRACTICE

Public demos, always pinned.

Start with either demo. They remain here when you filter or sort.

Free account: sign in to practice the marked Free questions.

Newest published questions; deterministic, never randomized.

Demo 1Number search with delayed feedback
Demo 2Design real-time device monitoring and control

Question bank filters

Reset filters

Question groups

Filter question bank by interview groupAll 269 Coding 162 System design 48 Behavioral 29 Hiring manager 16 Project deep dive 14

Topics

Independent and not affiliated with the companies represented. Collected material does not predict future interview questions.

255 questions

Sort question bankSortFree first Newest Oldest AI ratio ↓AI ratio ↑

Question bank results

Every question has a public brief with the task, constraints, and follow-ups. Labels describe interactive practice access.

Public brief

Topic tags
Python
Storage & state
Reliability

OpenAICoding

Python codingWebAssembly + public tests2 versions

Reported interview periodNov 2025–Aug 2026

Persist a string map across restart

Implement a durable string-to-string key-value store using the provided FileSystem abstraction.

Public brief

Topic tags
Python
Storage & state

OpenAICoding

Python codingWebAssembly + public tests

Reported interview periodDec 2025–Aug 2026

Turn-Based Creature Battle

Implement a deterministic battle between two teams of creatures. Team A acts first. On each turn, the first living creature on the active team attacks the first living creature on the opposing team.

New public brief

Topic tags
Python
Data structures

AnthropicCoding

Python codingWebAssembly + public tests

Reported interview periodJul 2025–Aug 2026

Reconstruct Stack Transitions from Samples and Events

Build utilities for converting between stack observations and call events. Stack frames are ordered outermost to innermost.

Public brief

Topic tags
Python

MetaCoding

Python codingWebAssembly + public tests

Reported interview periodJan 2026–Aug 2026

Leftmost local maximum in a numeric sequence

Given a sequence whose adjacent values never tie, locate the first element that rises above every neighbor it has. Report that element's position.

Public brief

Topic tags
Python

MetaCoding

Python codingWebAssembly + public tests

Reported interview periodMar 2026–Aug 2026

Count contiguous runs with a target sum

Given an integer array and a target, count the non-empty contiguous runs whose values add up to that target. Values may be zero or negative, so a qualifying run can begin and end anywhere.

Public brief

Topic tags
Python
Data structures
Storage & state
Networking

OpenAICoding

Python codingWebAssembly + public tests

Reported interview periodJan 2026–Aug 2026

Follow graph with snapshots

Implement a directed social graph with add_user, follow, and create_snapshot.

Public brief

Topic tags
Python
Algorithms
Security

DatabricksCoding

Python codingWebAssembly + public tests

Reported interview periodApr 2026–Aug 2026

Delete a Position from Interval Coverage

Public brief

Topic tags
Python
Concurrency
Reliability

DatabricksCoding

Talk-through codingNo local grader

Reported interview periodNov 2025–Aug 2026

Durable Multi-Threaded Log Writer

Implement a log writer that accepts binary records from multiple threads and durably appends each complete record to a file.

Public brief

Topic tags
Python
Concurrency

DatabricksCoding

Python codingWebAssembly + public tests

Reported interview periodNov 2025–Aug 2026

Hybrid Run-Length and Bit-Packing Codec

Implement a reversible encoder for an integer sequence using run-length blocks for long runs and fixed-size literal blocks otherwise.

Public brief

Topic tags
Algorithms
Data structures

OpenAISystem designAI-generated

Discussion / designResponse prompt

Reported interview periodJan 2026–Aug 2026

Crossword Constraint Solver

Given a crossword grid, a set of across/down clue entries, and candidate words for each clue, fill the grid so every entry uses one candidate and all crossing letters agree.

New public brief

Topic tags
Python
Concurrency
ML / AI

AnthropicCodingHalf AI-generated

Python codingWebAssembly + public tests

Reported interview periodSep 2025–Aug 2026

Build a Tool-Using Research Agent

New public brief

Topic tags
Python
Data structures
Storage & state

AnthropicCoding

Talk-through codingNo local grader

Reported interview periodApr 2026–Aug 2026

Convert Sampled Call Stacks into Events

A profiler periodically records the active call stack. Convert these snapshots into the minimal sequence of call-start and call-end events.

Public brief

Topic tags
Python
Networking
Security

DatabricksCoding

Python codingWebAssembly + public tests

Reported interview periodNov 2025–Aug 2026

Ordered IPv4 Firewall Rules

Implement an ordered firewall policy for IPv4 addresses.

Public brief

Topic tags
ML / AI
Product design
Communication

OpenAISystem designAI-generated

Discussion / designResponse prompt

Reported interview periodAug 2026

Design an interactive model-backed product feature

Public brief

Topic tags
Communication

OpenAIBehavioralAI-generated

Discussion / designResponse prompt

Reported interview periodDec 2025–Aug 2026

Recruiter screen: motivation, fit, and career narrative

Public brief

OpenAISystem designAI-generated

Discussion / designResponse prompt

Reported interview periodAug 2026

Engineering Management Fundamentals

Describe how you operate an engineering team and support both delivery and people growth. Use specific examples rather than generic management principles.

Public brief

Topic tags
Python
ML / AI
Product design

AnthropicCodingHalf AI-generated

Python codingWebAssembly + public tests

Reported interview periodSep 2025–Aug 2026

Implement Batched Scaled Dot-Product Attention

Implement attention for batched multi-head NumPy arrays on CPU. Inputs are finite arrays with the same floating-point dtype; dimensions are positive.

Public brief

Topic tags
Python
Data structures
Concurrency

OpenAICoding

Python codingWebAssembly + public tests2 versions

Reported interview periodDec 2025–Aug 2026

First-fit heap blocks

Implement a memory allocator over a fixed contiguous address range. Maintain free blocks in a doubly linked list ordered by starting address.

Public brief

Topic tags
Python

AnthropicCoding

Python codingWebAssembly + public tests

Reported interview periodOct 2025–Aug 2026

Build and Scale an Image-Transformation Pipeline

Public brief

Topic tags
Python
Data structures

AnthropicCoding

Talk-through codingNo local grader

Reported interview periodSep 2025–Aug 2026

Find Duplicate Files Efficiently

Walk a directory tree and return groups of files whose complete byte contents are identical.

Public brief

Topic tags
Python
Data structures
Storage & state

OpenAICoding

Python codingWebAssembly + public tests

Reported interview periodNov 2025–Aug 2026

Iterator that can pause and resume

Implement ResumableListIterator, an iterator that can save and restore its progress.

Public brief

Topic tags
Python
Data / SQL

OpenAICoding

Python codingWebAssembly + public tests

Reported interview periodApr 2026–Aug 2026

Online softmax entropy

Implement numerically stable softmax, log-softmax, and entropy for a vector of logits. The standard entropy is -Σ p log p.

Public brief

OpenAISystem design

Discussion / designResponse prompt

Reported interview periodAug 2026

Token-level streaming differ with rollback

Design an incremental processor for a token stream whose producer may append new tokens or roll back previously emitted tokens.

Public brief

Topic tags
Python
Algorithms
Concurrency
Distributed systems
ML / AI

OpenAICodingHalf AI-generated

Python codingWebAssembly + public tests

Reported interview periodJul 2026–Aug 2026

Debug sharded matrix-multiplication gradients

Implement forward and backward matrix multiplication across world workers.

Previous7 of 11Next

Browse all original practice

Build a token-budget batcher
Sliding-window rate limiter
Interval calendar
Nested dictionary transforms
Command-log store
Dependency-ready scheduler
Streaming metric window
Evaluation slice aggregator
Build a reliable job runner
Build one adaptable decision story
Defend your role fit
Prepare one project deep dive
Design an evaluation artifact registry
Defend a reversible launch decision
Negotiate an ambiguous role scope
Trace a project from claim to evidence
Choose between two urgent priorities
Resolve a cross-functional disagreement
Own a consequential failure
Increase a team’s leverage
Warm a cold cache without a thundering herd
Ship a 500 GB checkpoint without a startup stampede
Make every document decision traceable to evidence
Share eight GPUs between two workloads with different promises
A mutable indexed sequence that must restart exactly
Trace a falling unit price to a rising bill
Keep a very long conversation usable while it streams
Design durable scheduled publishing
Make speculative code edits safe to accept
Recover a grouping rule from two-word phrases

CONTENT BOUNDARY