stripe staff machine learning engineer interview experience and tips
Interview Experience
Preliminary round：
ML design - typical fraud detection problem.
ML
coding instead of classification problem，the learning problem has two regression tasks.
Things to note for ML coding
Among regression labels, one set is negative values, the other set is positive. The range of two labels differs quite a bit, so normalization is needed after flipping the signs if learning both tasks at the same time.
Full Details
Preliminary round：
ML design - typical fraud detection problem.
ML
coding instead of classification problem，the learning problem has two regression tasks.
Things to note for ML coding
Among regression labels, one set is negative values, the other set is positive. The range of two labels differs quite a bit, so normalization is needed after flipping the signs if learning both tasks at the same time.