Candidates are usually good at listing features for the given problem. It’s important to think about the problem and properly categorise the features, rather than listing generic features that may not be applicable here.

Two main things to consider in this step are:

• Structured approach to identifying and designing problem-appropriate features. 
• Feature extraction techniques.

The interviewer is usually looking for a comprehensive list of features that are applicable to the given problem and details on the features, how to extract these features, etc.

It’s important to have a logical grouping for the features, and do some analysis on which features might be important in this scenario. The candidate should also discuss how would these features be used by the model, talk about feature extraction techniques, propose multiple options wherever applicable and do some trade-off analysis.