1. Avoiding Interdisciplinary Bias and Discrimination Machine learning, as a discipline with a large breadth of crossover and a deep integration of disciplines, has interviewers from all kinds of backgrounds. I would recommend making sure not to presuppose that the interviewer has a unique knowledge base, whether it's math, statistics, physics, computers, or any other discipline. For example, machine learning actually overlaps with statistical learning for the most part, and even has a lot of relevance to optimization (e.g. operations research) and mathematics (linear algebra, etc.). And people in different fields may describe the same thing similarly but not the same, or even call it differently. As a simple example, statistics calls variables predictors while machine learning tends to call them features. I've heard of many interviewers rejecting candidates just because they can't quite use the terminology he understands, and I think this is silly. In the case of our team, my boss is a PhD and professor of statistics (econometrics biased) and I come from a pure computer science background. He loves modeling in R while I'm only good at Python and C++. But it is this difference that allows us to work better together. he likes to use various density estimation or direct fitting of distributions in unsupervised learning, while I can introduce him to many popular algorithms in the direction of machine learning, such as Isolation Forests. similarly, Python and R have their own specialties. Similarly, both Python and R have their own areas of expertise, for example, Python is much more difficult to do time series analysis than R, because R has a very mature package. therefore, we should not easily dismiss an interviewer just because of the different areas, different names, different programming languages, or different ways of interpreting the model. In the long run, our thinking will become narrower and narrower, while a certain degree of tolerance can expand the thinking.

2. The breadth test delineates the interviewer's knowledge of machine learning projects, which generally involve a series of processes such as data processing, modeling, evaluation, visualization, and deployment, and we expect the interviewer to have a basic understanding of each of these steps. Because of its broad scope, we want to first understand the scope of an interviewer's knowledge in a short period of time. There are many basic but classic questions that can be used to understand the qualities of an interviewer.