Interview question asked to Data Scientists interviewing at Bill.com, Scribd, Google and other companies. Original question asked: How do you prepare a specialized dataset for LLM fine-tuning to ensure the model adopts the required domain-specific tone?.