Many people learn Deep Learning by using frameworks.
Fewer learn it by building the engine themselves.
I’ve made public my teaching repository:
🔗 Neural Networks from Scratch (Jupyter Notebook Series)
This repository walks step-by-step through implementing a neural network using only NumPy — the way you would explain it on a whiteboard, but executable.
You’ll build everything manually:
- neurons & layers
- activation functions
- loss functions & gradients
- backpropagation
- optimization
- regularization & dropout
- regression & classification models
Originally created as graduate-level teaching material for Deep Learning courses, but it works equally well for self-learners who want intuition instead of “magic”.
If you’ve ever trained a model and thought:
I know it works… but why does it work?
This is for you.
Feedback, suggestions, and contributions are welcome ⭐