Different layers need different learning rates because lower layers contain stable coding fundamentals that shouldn’t change much, while upper layers hold task-specific logic that must adapt quickly to the new problem.

Hiring Assessment (screening quiz), two rounds with five interviews (did not pass first). Both contained two case study-esque questions. The first was about applied statistics and the second was about data intuition. I was rejected based on a "mixed" performance in the applied statistics interview, but was told I did well in the data intuition interview (reading graphs, charts and tables, analyzing them for meaning, discussing next experimental steps, other data science directions to move in based on what the graphs and said, etc.).

coding, we most often think that ML or Data science don't need dsa but it is actually essential i got a coding question and a puzzle and yeah coding questions were at the level like leetcode medium to difficult