Amazon Machine Learning Engineer Interview Questions

I would use a rolling window over the numeric column and then take the mean. In Pandas, that is usually a rolling(7).mean() pattern, with attention to the first few rows where a full window is not available. The reason I choose this approach is that it is clear, efficient, and easy to verify. I would also think about whether the data needs to be sorted by time before computing the window. If the interviewer wants more detail, I would mention how I would handle missing values or irregular intervals.

Normalization usually means scaling features to a common range so no single feature dominates because of raw magnitude. It matters most for distance-based methods like KNN, clustering, or models that are sensitive to feature scale. The reason I choose this explanation is that normalization is often confused with standardization, so I would make that distinction if asked. I would also note that tree-based models usually care less about scaling than linear or distance-based methods. If the interviewer wants more detail, I would give a small example with age and income features.

I would first identify the KPIs with stakeholders so the dashboard shows the numbers they actually use. Then I would define the data model, build the transformation layer, create the visualizations, and set up a refresh schedule. The reason I choose this approach is that a dashboard is only useful if the definitions are trusted and consistent. I would also think about access control, drill-downs, and whether the dashboard should support near-real-time or batch updates. If the interviewer wants more detail, I would explain how I would validate the numbers against source systems.

I would describe the situation, the stakeholder's concern, and how I aligned the technical work with the business need. I would focus on communication, trade-offs, and how I kept the project moving without ignoring the concern. The reason I choose this style is that Amazon values ownership and clear decision-making under pressure. I would also mention how I used data or prototypes to make the discussion concrete. If the interviewer wants more detail, I would explain what changed because of that conversation.

I would explain what was due, what was at risk, and how I prioritized the work. Then I would show how I broke the problem into pieces, removed unnecessary work, and kept stakeholders informed. The reason I choose this format is that Amazon interviewers often want to see ownership and execution under pressure. I would also mention any trade-off I made to hit the deadline and whether I would do anything differently next time. If the interviewer asks, I would connect the result to the user or business impact.