Katkat

Topic

ML System Design interview questions

332 ML System Design questions from 100 companies, with hints, runnable code and solution walkthroughs.

Questions332

13 easy187 medium132 hard

Your progress0/332

ML System Design interview questions by the numbers

Last updated

Katkat has 332 ML System Design interview questions from 100 companies; the most come from OpenAI (7%), Meta (7%) and Anthropic (6%). By difficulty: 4% easy, 56% medium and 40% hard. The most common round is Onsite (56%). Frequent concepts include System Design, ML System Design, Scalability and LLM.

Companies (100)

Difficulty

  • Easy4% · 13
  • Medium56% · 187
  • Hard40% · 132

Interview rounds

  • Onsite56% · 186
  • Technical Screen42% · 138
  • Online Assessment2% · 6
  • HR Screen1% · 2

Roles

  • Machine Learning Engineer54% · 178
  • Software Engineer36% · 121
  • Applied Scientist4% · 13
  • Data Scientist4% · 13
  • Member of Technical Staff, ML Systems1% · 2

Frequent concepts

  • System Design17% · 56
  • ML System Design10% · 32
  • Scalability8% · 28
  • LLM8% · 26
  • RAG8% · 26
  • Ranking6% · 20

Most mentioned in interview experiences

  1. 1.Design a Spam Review Workflow with Classifiers, AI Agents, and Human Labelers2 reports
  2. 2.Add an LLM-backed recommendation API to an existing service with RAG and prompting1 report
  3. 3.Applying LLMs to Autonomous Driving Despite Hallucination and Real-Time Limits1 report
  4. 4.Architect an asynchronous RL post-training system1 report
  5. 5.Build a Mistral-powered RAG agent1 report
  6. 6.Build AI chat for spreadsheets1 report
  7. 7.Choose File Storage Tiers Using Access Predictions1 report
  8. 8.Compare Single-Agent and Multi-Agent Designs1 report
  9. 9.Data quality, lineage, monitoring and backfill design for an ML feature store1 report
  10. 10.Describe model-to-GPU execution pipeline1 report

Counted from 82 candidate interview experiences on Katkat. How we collect and count

Showing 331–332 of 332 results

  • Design a Retrieval-Augmented Generation (RAG) SystemMedium

    This question evaluates understanding and engineering competency for retrieval-augmented generation (RAG) systems, covering offline indexing and online query planes, retrieval quality, prompt assembly, grounded generation with citations, and production monitoring; the domain is ML System Design.

    xAISoftware EngineerML System DesignOnsite
  • Implement a trie-based tokenizerHard

    This question evaluates a candidate's competency in designing and implementing a production-grade subword tokenizer, covering trie-based longest-prefix matching, Unicode-aware text processing, normalization and casing impacts, whitespace/punctuation rules, fallback strategies, performance and memory trade-offs, versioning, and testing for edge cases. It is commonly asked in ML System Design interviews to assess practical implementation skills and conceptual trade-off reasoning for deterministic, scalable LLM preprocessing, and tests both practical application and conceptual understanding within the ML System Design domain.

    xAIMachine Learning EngineerML System DesignTechnical Screen

More topics