The first step is a one hour interview with a scientist at the company. They walk through your resume and ask about relevant experiences and basic machine learning questions. Nothing tricky, they just want to gauge how well you understand machine learning concepts and whether you can explain the basic algorithms.
Then the second step is a take home coding challenge. It consists of building a machine learning pipeline for a nlp task. There's some tricky parts about it, but it's very straight forward.
Interview questions [1]
Question 1
1. Describe the bias variance tradeoff.
2. How would you measure the performance of a classifier?
3. What is an example of a problem where you would favor recall over precision and vice versa.?
4. Describe the main idea behind boosting along with some well known algorithms such as adaboost and xgboost.
5. Describe several options for embedding text data into a feature along with pros and cons of each.
6. Describe your favorite sorting algorithm