LLM Inference & Serving Q&A

Part of the Interview Q&A track on AI-ML Companion.

24 scenarios from inference and LLM-infra loops - the bandwidth arithmetic, engine internals, quantisation, multi-GPU topology, production debugging and capacity