Phone screen was split into three components with a current ML Software Engineer. First part was just introductions and backgrounds for both interviewer and candidate.
The second component, involved a coding question similar to “Implement rand10 using rand7”:
https://leetcode.com/problems/implement-rand10-using-rand7/
However, the given random number generator wasn’t sampled from a uniform distribution.
The final aspect of the call focused on ML System Design question. It focused on a LinkedIn related classification problem. Candidate is intended to talk interviewer through designing dataset, features, model, etc. Interviewer will dive deeper into these topics: adding additional constraints, asking for further details, changes to original problem, etc.
Phone call ended with any questions candidate may have about LinkedIn, including projects, work-life, etc.