I've adopted a similar attitude as you here when it comes to past machine learning jobs, and discussion of detail. What ends up being your bright shiny line that you don't cross? I tend to just not talk about the specific feature engineering, being relatively upfront about such basic things as "I used a random forest".