AI and ML engineer interview questions by company
262 questions candidates reported being asked at 58 AI labs and big tech companies, each with a model answer.
OpenAI
AI lab · 9 questions
Context & KV cacheHallucinationContext selectionRAGAnthropic
AI lab · 10 questions
AttentionLatency & throughputWhere the time goesCost per requestMistral AI
AI lab · 16 questions
AttentionSamplingContext & KV cacheWhere the time goesMeta
Big tech · 9 questions
AttentionPrompt injectionFine-tuning & alignmentAgentsMicrosoft
Big tech · 11 questions
EmbeddingsSamplingHallucinationStructured outputsGoogle
Big tech · 11 questions
TokenizationSamplingCost per requestBenchmarkingAmazon
Big tech · 13 questions
TokenizationAttentionContext & KV cacheLatency & throughputApple
Big tech · 6 questions
EmbeddingsAttentionHallucinationCost per requestNVIDIA
Big tech · 6 questions
Latency & throughputWhere the time goesQuantizationBenchmarkingNetflix
Big tech · 7 questions
TokenizationAttentionContext & KV cacheTrainingByteDance
Big tech · 8 questions
AttentionBenchmarkingRAGFine-tuning & alignmentScale AI
AI startup · 7 questions
EmbeddingsSamplingStructured outputsPrompt injectionCohere
AI startup · 7 questions
SamplingHallucinationRAGFine-tuning & alignmentDatabricks
AI startup · 4 questions
Prompt injectionRAGPerplexity
AI startup · 3 questions
Tools & constraintsTrainingML system designAdobe
Big tech · 4 questions
EmbeddingsAttentionWhere the time goesTrainingAirbnb
Big tech · 4 questions
RAGML system designApplied Intuition
AI startup · 3 questions
AttentionContext & KV cacheTrainingAtlassian
Big tech · 3 questions
RAGML system designAurora
AI startup · 1 question
HallucinationC3 AI
AI startup · 4 questions
EmbeddingsRAGTrainingCapital One
Big tech · 4 questions
Success criteriaTrainingML system designCitadel
Big tech · 2 questions
BenchmarkingTrainingCoinbase
Big tech · 1 question
TrainingDatadog
Big tech · 3 questions
Context & KV cacheAgentsTrainingDecagon
AI startup · 3 questions
Latency & throughputPrompt injectionRAGDoorDash
Big tech · 3 questions
TrainingML system designeBay
Big tech · 2 questions
Latency & throughputML system designFigma
Big tech · 4 questions
TokenizationAI judgesFine-tuning & alignmentML system designGlean
AI startup · 2 questions
TokenizationEmbeddingsGrammarly
Big tech · 1 question
Tools & constraintsHarvey
AI startup · 6 questions
EmbeddingsRAGAgentsIBM
Big tech · 1 question
ML system designJane Street
Big tech · 1 question
TrainingLinkedIn
Big tech · 5 questions
EmbeddingsTrainingML system designLuma AI
AI startup · 3 questions
Regression testsTrainingMultimodalLyft
Big tech · 4 questions
Success criteriaRAGFine-tuning & alignmentTrainingMercor
AI startup · 5 questions
Context & KV cacheContext selectionReference scoringFine-tuning & alignmentNuro
AI startup · 2 questions
Latency & throughputTrainingOracle
Big tech · 1 question
HallucinationPayPal
Big tech · 4 questions
Success criteriaML system designPinterest
Big tech · 6 questions
EmbeddingsTrainingML system designMultimodalReddit
Big tech · 3 questions
ML system designRoblox
Big tech · 3 questions
ML system designMultimodalShopify
Big tech · 2 questions
ML system designSierra
AI startup · 2 questions
Latency & throughputContext selectionSnap
Big tech · 8 questions
EmbeddingsAttentionTrainingML system designSnowflake
Big tech · 2 questions
Success criteriaAgentsSpotify
Big tech · 2 questions
TrainingML system designTesla
Big tech · 4 questions
AttentionTrainingTikTok
Big tech · 10 questions
AttentionContext & KV cacheReference scoringFine-tuning & alignmentUber
Big tech · 5 questions
AttentionError analysisTrainingML system designWalmart
Big tech · 2 questions
AgentsML system designWaymo
Big tech · 6 questions
Latency & throughputQuantizationAI judgesComparing systemsWorkday
Big tech · 1 question
RAGxAI
AI lab · 1 question
TrainingYahoo
Big tech · 1 question
TrainingZoox
AI startup · 1 question
Benchmarking
Each question comes from a public interview write-up or video that names the company. Interviews vary by role, team and year. Not affiliated with any company listed.