Prompt engineering for cloud specialists

- [Presenter] Prompt Engineering is the art of creating effective prompts that guide AI to deliver the desired outcomes. Effective prompt construction often requires intuition, experience, and iterations. Let's look at some key concepts and tips in prompt engineering as a cloud specialist. Tip number one, context is king. Define role and goal. This sets the context, shapes the AI's persona and voice, provides a frame of reference for knowledge, helps the AI prioritize relevant information and tells the AI clearly what you are expecting. For example, specify in the prompt that you are an experienced Azure Cloud architect tasked to architect a retrieval augmented generation application built entirely in a serverless environment on Microsoft Azure to ensure maximum scalability and minimal operational overhead. With this sample, we immediately tell the AI to activate its knowledge and reasoning abilities through the eyes of this specific individual. Given that there are so many platforms and approaches in cloud computing, the sample lets the AI focus on the Azure stack and on a serverless approach. Tip number two, decompose and question. Break down and ask why and how. Large and complex cloud problems are hard for AI to tackle in a single prompt. It is better to decompose tasks into smaller and manageable steps. In addition, asking why and how will get the AI to explain underlying mechanisms and reasoning within the responses and helps you debug its reasoning if needed. Tip number three, guided examples and format, show then tell. Examples clarify your needs, especially for code or configurations. We need to specify the output formats for structured results. Showing the AI what you want will lead to more precision. Tip number four, ground in official documents. Link and name drop. Grounding the AI in the relevant cloud provider's documentation can significantly improve the accuracy and relevance of responses. If the AI allows, provide links to specific documentation pages or refer to specific services and features by their official names. And finally, tip number five, iterate and refine. Prompting is a process. Don't expect perfect results on the first try, even if you apply everything you have learned on prompt engineering. Based on the AI's response, you should ask follow up questions, request deeper dives, or adjust your prompts. Through practice and continuous improvement, you will eventually develop other techniques that may help you in your work. At the end of this course, I've provided you with a list of sample prompts that you can use to guide you in your daily work.