Question bank

4,310 interview questions, answered.

Reference answers, what the interviewer is really testing, how it is graded, and the follow-ups that come next.

Easy 6Medium 24Hard 33
Filters · onDifficulty, topic, company
Clear all

24 questions

  1. 1Implement a BPE tokenizer from scratchMediumCodingSystem design· OpenAI
  2. 2Implement an async task queue with a concurrency limitMediumCodingSystem design· Neon
  3. 3Implement beam search decodingMediumCodingSystem design· PAPerplexity AI
  4. 4Generate outputs for images and pipelines (Anthropic)MediumCodingSystem design· Anthropic
  5. 5Build a rate-limited async LLM API clientMediumCodingSystem design· Cohere
  6. 6Cohere infrastructure utilities (5-in-1 take-home)MediumCodingSystem design· Cohere
  7. 7Compute entropy and implement 1-NN (OpenAI)MediumCodingSystem design· OpenAI
  8. 8Cursor 8-hour paid take-home build sessionMediumCodingSystem design· Cursor
  9. 9Detect and handle hallucinations in LLM outputsMediumCodingSystem design· AEAI Engineer / Eval roles
  10. 10Implement scaled dot-product self-attentionMediumCodingSystem design
  11. 11Implement temperature, top-k, and top-p samplingMediumCodingSystem design
  12. 12In-memory file system coding challenge (Perplexity)MediumCodingSystem design· PAPerplexity AI
  13. 13Mistral 60-minute live coding (algorithm + practical AI utility)MediumCodingSystem design· Mistral AI
  14. 14Write a training loop with gradient accumulation and mixed precisionMediumCodingSystem design
  15. 15Design an AI tutoring platform with RAG-based Q&AMediumCodingSystem design
  16. 16Implement cosine-similarity search from scratchMediumCodingSystem design
  17. 17Build a simple AI agent with tool useMediumCodingSystem design
  18. 18Build a sliding-window + summary conversation memoryMediumCodingSystem design
  19. 19Build a small RAG app (chunking, embedding, retrieval, answer)MediumCodingSystem design
  20. 20Implement a simple re-ranker for search resultsMediumCodingSystem design
  21. 21Implement a streaming response handler for an LLM APIMediumCodingSystem design
  22. 22Implement an eval pipeline with LLM-as-a-judgeMediumCodingSystem design
  23. 23Implement semantic search using embeddingsMediumCodingSystem design
  24. 24Write a chunking strategy module (fixed-size / recursive / semantic)MediumCodingSystem design