Your question is Fine-Tuning vs Prompt Engineering. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
What are the primary challenges when deploying a fine-tuned model versus using a prompt-engineering approach?