Welcome to AIby{Examples} ๐
"What I cannot create, I do not understand."
โ Richard Feynman
Traditional AI education is often polarized:
- Shallow High-Level Overviews: Black-box tutorials that import a library (
model.fit()) without explaining the mechanics under the hood. - Dense Academic Treatises: Heavy mathematical proofs and walls of equations that obscure practical intuition and real-world engineering constraints.
AIby{Examples} bridges this gap through active, first-principles construction. Every topic is taught through concrete examples, hand-drawn visual sketchnotes, step-by-step mathematical derivations, and runnable, zero-magic Python code.
How every guide is builtโ
Each topic in our curriculum moves through the same three-part arc:
| Stage | What you get |
|---|---|
| ๐ก See it | A clear mental model and visual sketchnote โ no syntax yet, just intuition for why the idea exists. |
| โก Run it | The smallest possible Python example (zero dependencies) plus the underlying math, derived step by step. |
| ๐ ๏ธ Ship it | A production-grade implementation, its common pitfalls, and hands-on challenges to test your understanding. |
๐งช Try It Yourself: In-Browser Playgroundโ
Experience our zero-install, in-browser Python execution (powered by WebAssembly/Pyodide) and interactive mental models directly below:
1. Dynamic Visual Analogy: Loss Landscapeโ
Adjust the stride length (learning rate ) slider below to see how steps oscillate or smoothly converge into the loss minimum:
Gradient Descent & Step Size
The Hiker in the Dense Fog: Imagine walking downhill in dense fog. You can only feel the local slope beneath your boots. Your stride length determines whether you smoothly descend into the valley or overshoot the path entirely.
Mathematical Reality: Parameter Updates: In neural network training, we calculate the slope (gradient) of the loss with respect to weights. The learning rate dictates the step magnitude along the negative gradient.
In real deep learning, loss surfaces have non-convex saddle points and billions of dimensions where 2D spatial intuition no longer applies.
2. Cell-Level Python Notebookโ
Run and edit Python code right in your browser. State is shared across cells!
What you'll learnโ
Our curriculum follows the full AI lifecycle, from first principles to production and beyond:
| # | Phase | Focus |
|---|---|---|
| 01 | Fundamentals | Math, statistics, and programming foundations for AI |
| 02 | Algorithms & Model Theory | Classical ML, neural networks, and deep learning architectures |
| 03 | Development | Data pipelines, model training, agents, and RAG systems |
| 04 | Optimization | Compression, quantization, and efficient inference |
| 05 | Deployment | System design, model serving, and MLOps |
| 06 | Monitoring | Observability, drift detection, and cost tracking |
| 07 | Refinement | Continual learning, experimentation, and debugging |
| 08 | Responsible AI | Safety, fairness, privacy, and interpretability |
| 09 | Research & Advanced | Emerging architectures and the AI research frontier |