What is Apache Kafka and Kafka Streams?
Machine Learning / AI
What is async checkpointing and how does it work?
How to ensure data security in RAG (PII, access rights)?
What is TRPO and the idea of trust region?
What is censoring and what types are there?
How often do new tasks appear and what is the workload?
What is class-aware vs class-agnostic NMS?
What is sMAPE and what are its problems?
What is VAE for anomaly detection?
Что такое Wav2Vec 2.0 и self-supervised pretraining?
What is a model inversion attack?
How to calculate sample size for an A/B test?
How to update the RAG index in incremental updates mode?
How does MMLU differ from MMLU-Pro?
Who is the customer and how are tasks formed?
What prompt engineering techniques have you applied (few-shot, role, CoT, ReAct)?
Explain the TF-IDF formula and the intuition behind each component.
What tasks might be expected in the next six months?
Which techniques are cheaper than human eval (LLM-as-a-judge, model-graded)?
What processes do you like in a team?