Python for AI — The 20% You Use in Every Notebook

You do not need to be a Python expert to do AI — you need a specific slice. Here it is.

The data structures you will live in

# list — ordered collection
scores = [88, 92, 79]
scores.append(95)

# dict — key/value (config, JSON, records)
student = {"name": "Priya", "cgpa": 8.9}
student["dept"] = "CSE"

# the AI workhorse: list comprehension
squared = [x**2 for x in scores]              # transform
passed  = [x for x in scores if x >= 80]      # filter

Functions and the shapes you pass around

def normalize(values):
    lo, hi = min(values), max(values)
    return [(v - lo) / (hi - lo) for v in values]   # scale to 0..1

normalize([10, 20, 30])   # [0.0, 0.5, 1.0]

The notebook workflow

AI work happens in Jupyter notebooks (or Google Colab — free, browser-based, free GPUs). Code runs in cells you can re-run independently, with plots and tables inline. It is a REPL built for experiments.

• Use Colab to start — no install, GPU for deep learning, just a Google account.
• Shift+Enter runs a cell. Variables persist between cells.
• Print shapes constantly: print(data.shape) — 90% of AI bugs are shape mismatches.

Already know JS? See our Python for JavaScript devs translation guide. Next: NumPy & Pandas, the real tools.