ML for AI Engineers

What you'll learn

1. 1

   Neural Networks in One Lesson

   Neurons, layers, activations and parameters: what's really inside the models you call.

   • Weighted sums, stacked
   • Forward pass of a layer
   • Count the parameters
   • Building blocks

   Take-home template: Neural Network from Scratch (NumPy)

2. 2

   How Models Learn: Gradient Descent

   Loss, gradients and learning rate: the engine behind every neural network.

   • Learning means reducing a number
   • Tune the learning rate
   • One gradient descent step
   • Read the training curve

   Take-home template: Gradient Descent from Scratch (NumPy)

3. 3

   Data Splits, Overfitting & Leakage

   Why a model can look perfect in development and fail in the real world.

   • Never grade on the homework
   • Diagnose it
   • Catch data leakage

   Take-home template: Leak-Free Dataset Splitter

4. 4

   Prompting vs RAG vs Fine-Tuning

   The decision every AI team faces, with a practical guide.

   • Three levers, in this order
   • Pick the lever
   • A common mistake
   • Build a fine-tuning dataset
   • You know the engine. Now build the car

   Take-home template: Fine-Tuning Data Kit