How to use this course

This course is a workshop, not a book you race through.

You will read short explanations. More importantly, you will predict, run, change, and repair small examples. Those actions turn an idea you recognise into a skill you can use.

Why this matters

Watching someone code can feel like learning. The real test comes later: can you explain the idea, change the code, and recover when it fails?

This lesson gives you a repeatable study method. It also helps you spend your time where depth matters most.

Think of the course as a guided expedition

A course layer is a group of lessons that builds one part of your AI engineering knowledge. Foundations sit underneath later application topics.

A depth tier tells you how far a lesson expects you to go:

• Build: create the idea yourself and explain its moving parts.
• Use: apply it correctly and compare its trade-offs.
• Know: recognise it, explain its purpose, and know when to revisit it.

See the whole course

The lower layers introduce language and tools used later. If a later lesson feels as though it skipped a step, follow its prerequisite link and return when the missing idea feels clear.

Follow the learning loop

Step 1: predict

Before running code, say what you expect. A prediction gives you something concrete to compare with the result.

Step 2: run

Select Run. The interactive Python cells use PyodidePyodideA build of the CPython interpreter compiled to WebAssembly, so real Python — numpy, pandas, scikit-learn and more — runs entirely inside your browser tab with no server. It is what executes the runnable cells in this course., which runs Python inside your browser.

Step 3: break

Change one value or remove one line. This is deliberate practice: focused practice designed to expose what you do and do not yet know.

Step 4: read the feedback

When Python cannot continue, it produces a traceback. A traceback is an error report that shows where execution stopped and names the type of error.

Step 5: tune and explain

Repair the example. Then explain the cause in one sentence. If you cannot explain it yet, compare your code with the worked example and try again.

Worked example

Suppose you want to study for six hours each week:

hours_per_week = 6
weeks = 4
practice_hours = hours_per_week * weeks
print(practice_hours)

Predict the result first. Python multiplies 6 by 4, stores 24 in practice_hours, and prints:

24

Now break the code by changing weeks to "four". Quotation marks turn it into text. Multiplying text by 6 repeats it rather than performing the calculation. The unexpected output teaches you that the shape of data matters.

Tune the program by changing the value back to the number 4. Then explain the lesson: numbers and text behave differently, even when they describe the same idea.

Run the code

def study_action(minutes):
    if minutes >= 20:
        return (
            "Run, break, and "
            "repair the example."
        )
    return "Review six flashcards."


print(study_action(25))
print(study_action(10))

Run the example below. Change minutes_today, predict the new message, and run it again.

Real-world AI engineering use

Common mistakes

Knowledge check

Choose an answer. Feedback appears immediately and explains why.

Practice activity

Lesson:
My prediction:
Change I will make:
How I will check the repair:
When I will practise:

Lesson:
  Python foundations
  for AI applications
My prediction:
  Blank messages will be removed.
Change I will make:
  Move return inside the loop.
How I will check the repair:
  Every useful message appears
  after the fix.
When I will practise:
  Tuesday, 19:00–19:20.
Fallback:
  Review the six lesson flashcards.

Key takeaways

Flashcards

One-page sketchnote summary

predict()
run()
change_one_thing()
observe()
repair()
explain()

Resources