Published: Jul 13, 2023

What makes a good Prompt Engineer

As a machine learning engineer with seven years of experience in practicing (and always learning) data science, I have observed that the qualities that constitute a good machine learning engineer or data scientist also define a competent prompt engineer.

1. Broad Exposure: A successful prompt engineer is one who has exposed themselves to a wide array of problems, and possesses, at minimum, an intuitive understanding of why certain solutions are effective. This vast array of experiences builds their toolbox of problem-solving strategies.

2. Process Efficiency: A prompt engineer should have a systematic approach that allows for quick establishment of a baseline. From there, they should be able to iteratively enhance their solutions, constantly linking new problems to previously encountered ones. This is about discerning which tools are appropriate for specific situations.

Lastly, remember to meticulously test edge cases. This will help you identify areas where your solution falls short. It’s advisable to trial your prompts on a diverse range of scenarios rather than relying on a single example. By doing so, you are more likely to identify potential areas where your prompt might be misunderstood by the model. There will always be cases that fail, when you find them, just celebrate that there is another you caught. 😅