AI ENGINEER FOUNDATIONS · STAGE 03
Machine-learning fundamentals
Train models and measure whether they generalize.
Learn the basic learning setups, how training and validation differ, and how evaluation connects to a real task. Build intuition for overfitting and model selection.
BEFORE YOU START
Prerequisites
- Basic statistics and data handling
LEARNING OBJECTIVES
What you should be able to do
- Distinguish common learning signals
- Use train, validation, and test splits responsibly
- Explain generalization and overfitting
WHAT YOU'LL COVER
Key concepts
SEE THE CONNECTIONS
A visual learning sequence
These are connected learning steps, not the only valid order. Revisit earlier ideas when an exercise reveals a gap.
Suggested learning sequence
Frame the task
Identify the prediction target and available feedback.
Why it matters: The target defines what a model can learn and what success means.
Train and validate
Fit on training examples and use validation data for decisions.
Why it matters: Performance on training data alone can be misleading.
Evaluate in context
Select metrics and data splits that reflect the deployment scenario.
Why it matters: A score is only useful when it matches the cost of errors.
Put it into practice
For a support-ticket classifier, compare precision and recall and make sure validation data is separated by time if future tickets will be predicted.
Practice tasks
- Compare supervised, unsupervised, and reinforcement learning.
- Draw training and validation learning curves.
- List two ways data leakage could occur.
How you'll know you're ready
- Can explain what signal a learning setup receives.
- Can describe how to detect overfitting without reusing the test set.