Initial screen with hiring manager, followed by an initial technical screen with the development lead. Both went well and covered general ML systems knowledge.

Then a "virtual onsite" set of panel interviews, with two interviewers per session. These included coding tests, system design, and general ML knowledge.

The first coding test involved splitting rows of a table or dataframe across multiple processes and shuffling the data. The high-level code was acceptable; it didn't need to be completely detailed.

Another coding test was finding the maximum depth of a binary tree.