NumPy

NumPy provides the array data structure and vectorized operations that nearly every other library in Python’s data and ML stack — Pandas, scikit-learn, PyTorch — builds on top of. Its core advantage over plain Python lists is speed: operations run as compiled C loops instead of interpreted Python, which matters enormously once you’re processing anything beyond toy-sized data.

You’ll rarely reach for NumPy in isolation once you’re doing real work — it’s the layer underneath the tools you actually call, whether or not you import it directly yourself.