Domanda di colloquio di e-Procurement Technologies

How would you serve an LLM model at scale? How do you deploy an LLM locally using FastAPI or Docker? What are the challenges of deploying an LLM in a cloud setup like AWS or GCP? How would you handle multiple requests coming from the same user to your LLM service? What would you do to ensure consistent responses while maintaining performance? Explain how a RAG system works? Which vector database would you use and how do you manage chunking and embeddings?