A recruiter call, a demanding multi-part coding screen or assessment, then a virtual onsite of coding, search-and-AI system design, LLM product design and a founder or VP round; three to five weeks overall.
25 questions
In-memory file system with files and foldersCodingImplement an in-memory file system with path-based file and folder access, plus follow-ups on per-directory limits, duplicate names and identical contents. Covers the bank's vault…
Last asked Oct 2026
Design durable execution for long-running agent queriesAI systemAn answer service runs each query in memory, but agentic queries can now take hours. Redesign it so a query keeps running and finishes after the browser disconnects, resumes after…
Last asked Sep 2026
Agent to-do list with task dependenciesCodingBuild the task tracker an AI agent uses to plan its work. The question comes in three parts. Each part builds on the one before, so keep earlier behavior working. A status is an…
Last asked May 2026
Infinite-scroll news feed in TypeScriptCodingUsing supplied mock data and fetch helpers, build a feed page of article cards in TypeScript that loads more as the reader scrolls. Show clean component structure, pagination…
Speed up a byte tokenizer and estimate token countsCodingYou are given a slow but correct reference tokenizer over bytes. Explain precisely what it does, then write a faster version that returns identical output. Finally, estimate the…
Last asked Jan 2026
Train and improve a binary classifier with scikit-learnCodingGiven a small labelled dataset, train a binary classifier with scikit-learn, report a sensible evaluation metric, and then improve it through feature handling, model choice or…
Last asked Dec 2025
Reservoir sampling and a check that it is uniformCodingPick k items uniformly at random from a stream of unknown length in one pass and O(k) memory. Then write a simulation that tests whether your picks are close to uniform, and…
Last asked Nov 2025
Stream deduplication with near-duplicate suppressionCodingPass through a stream of strings, keeping only the first copy of each. Then also drop strings that are near-duplicates of something already kept, above a similarity threshold, and…
Last asked Nov 2025
Truncate model output at the first stop sequenceCodingImplement a function that returns the part of a string that comes before the earliest stop word. - text is any string. It may be empty. - stopwords is a list of non-empty strings…
Possibly retired
Batching service around an embedding modelCodingPut a service around an embedding model: many callers send texts at once, the service groups them into batches for the model, returns each caller its vectors, and stops without…
Possibly retired
Build your story bankBehavioralPrepare 8 to 15 real stories, each tagged with the signals it shows (ownership, conflict, failure, ambiguity, influence, impact, mentorship, quality), so that any behavioral…
Possibly retired
Find a way out of a restricted map zoneCodingGiven a starting latitude and longitude and a function that says whether a point is inside a premium zone, return a path that leaves the zone, or a nearby point outside it…
Possibly retired
Design a feed recommendation systemML designRank the items in each user's home feed so they engage more, choosing from millions of candidates within a couple of hundred milliseconds.
Possibly retired
Design a top-K popular items systemSystem designReport the K most viewed or played items over sliding windows (last minute, hour, day) from a firehose of billions of events per day.
Possibly retired
Greedy and beam search decoding for a language modelCodingYou are writing the decoding step for a language model. Decoding turns the model's per-step scores into a sequence of token ids. Implement greedydecode(scores: listlistfloat) -…
Possibly retired
Debug an overloaded Kubernetes clusterSystem designAn infrastructure round: a service running on Kubernetes is overloaded. Say which metrics you want to see first; the interviewer draws them for you, and you narrow the cause down…
Possibly retired
Design Mint.comSystem designLink users' bank and card accounts, pull and categorize transactions daily, and show budgets and spending trends, keeping credentials safe.
Possibly retired
LLM provider pool with failoverCodingWrite a client that sends completion requests to a pool of external LLM providers. It should be able to switch between providers, and when one goes down, move traffic to another…
Possibly retired
GPU credit ledger with out-of-order eventsCodingBuild an in-memory ledger for GPU credits. Credit arrives as grants: each one has a unique id, an amount, and a time window in which it can be spent. Grants may overlap in time…
Apply ordered token merges to a stringCodingMany language models read text as a list of token ids. In this scheme, every token after the single characters is built from two earlier tokens, and turning text into ids means…
Design a RAG system over company docsAI systemEmployees should be able to ask questions about internal documents (Drive, Confluence, Slack, GitHub) and get grounded answers with citations.
Design an answer engine with citationsAI systemDesign an answer engine: a user asks a question in plain language and gets a direct, current answer written by a model, with numbered citations to the web pages it used, streaming…
Design typeahead suggestionsSystem designSuggest completions as the user types into a search box, within tens of milliseconds, ranked by popularity and fresh with trending queries.
Question types
Perplexity interview process and questions · Land the Lab
