Interview Practice Questions

Do you have an on-call rotation? How is service monitoring organized?

How does the IVF (Inverted File Index) work in FAISS?

Model training takes 12 hours. How would you identify the bottleneck and speed it up?

How would you keep ML model inference latency below 100 ms?

Describe your experience with MLOps and model deployment.

How did you handle the embedding dimensionality limit of 256 in OpenSearch?

Have you written production code? How automated was the deployment process?

What tests did you write for the ML model or assistant, and how did you evaluate quality?

Will precomputed scores be stored in a cache, or will inference run at request time?

Is the solution you are describing already complex, or is it still a simple one?

Estimate how long this solution would take to implement with a team of three: two ML engineers and one backend engineer.

What problems do you think we need to anticipate at the architecture level?

Design a chatbot system that should handle 1000 active users concurrently and respond to queries within 2 seconds. What architecture would you propose and which technologies would you use?

Design a document classification pipeline that should process 1 million documents per week and achieve at least 85% accuracy. What stages would you include in the pipeline and which algorithms would you use?

Design a named entity recognition (NER) system that should process 10,000 texts per day and have a response time of less than 1 second. How will you ensure the scalability and reliability of this system?

Design a search system that should support 500 QPS and provide both semantic and keyword search capabilities. What approach would you take to integrate these two types of search?

Design a text summarization service that should process 100,000 texts per month and return summaries no longer than 150 words. What will be your architecture and technologies?

Design a retrieval-augmented generation (RAG) system that should handle 2000 user queries per hour and provide answer quality of at least 90%. How would you organize the data storage and models?

Design a content moderation system that should process 1 million messages per day and achieve a false positive rate of less than 5%. What strategies and algorithms would you apply?

Design a system for serving large language models (LLM) that supports 100 QPS and has a response time of no more than 500 ms. What technologies and approaches would you use to meet these requirements?