Interview essentials for NumPy

NumPy interviews test ndarray mechanics, broadcasting, memory views, vectorization, and linear algebra basics—often in Python DS/ML loops after Python and Data Science fundamentals.

High-frequency topics

• Shape, dtype, axis semantics
• Broadcasting rules and failure cases
• View vs copy; fancy vs slice indexing
• Boolean masking and np.where
• Matrix multiply vs element-wise
• NaN handling and nan-aware aggregates

Whiteboard pattern

1. State input shapes and dtypes
2. Draw axis direction for reductions
3. Explain memory sharing for reshape/transpose
4. Propose vectorized solution before loops

Sample verbal answer

“I'd use boolean indexing to filter, then nanmean on axis=0 for column summaries—avoiding Python loops and handling missing values.”

Important interview questions and answers

1. Q: Broadcast (3,1) + (3,)?
   A: Result shape (3,3)—column vector broadcast across rows.
2. Q: Why vectorize?
   A: Python loop per element vs C batch—critical at million-row scale.

Self-check

1. Name four common NumPy interview topics.
2. Explain view vs copy in one sentence.
3. How do you ignore NaN in mean?

Tip: Practice explaining broadcasting with a 2D sketch.