From NumPy and machine learning to neural networks, transformers and LLM apps, project by project
A guided AI engineering course in 16 chapters, from NumPy arrays, pandas and evaluation metrics through scikit-learn, gradient descent, neural networks and PyTorch to NLP, embeddings, transformers, RAG, scaling and MLOps. Every project shows its code and its exact output.
Builders who want the AI engineering stack explained through runnable projects instead of abstract jargon.
A project-based path from data foundations to modern AI engineering: 209 practice projects with code, exact output, quizzes and answer keys.
AI Systems Craft is designed as a practical bridge from ordinary programming curiosity into real AI system building. It does not begin by assuming advanced mathematics, university machine learning, or prior production experience. The book starts with visible data, small Python programs, tables, arrays, and metrics, then turns each new AI idea into a runnable experiment you can inspect. The goal is not to memorize buzzwords; the goal is to understand what the data looks like, what the model is being asked to do, how the output is measured, and how to make one safe improvement at a time.
Read the data shape, name the transformation, predict the metric or output, then change one detail safely.
16 chapters in three parts, from NumPy fundamentals to scaling, compute, cost and MLOps
209 practice projects, each with runnable code and its exact output
Evaluation metrics, scikit-learn, gradient descent, neural networks from scratch, PyTorch, vision, NLP, embeddings, transformers and RAG
A Logic Lens and a Try It for every project, plus a quiz and answer key in every chapter
Look inside
Pages from the PDF.
Real pages from the book. Select a page to open it at full size.
How to use the courseEvery chapter follows the same rhythm: the idea, a runnable project with its output, common mistakes, experiments and a quiz.Sixteen chaptersBeginner, intermediate and advanced parts run from NumPy arrays to scaling, cost and MLOps.The idea behind itChapter 1 opens with why shape, dtype and indexing matter, shown in code with its output.Worked examplesExamples are solved step by step before the chapter's first project begins.Later chaptersChapter 14 builds transformers and attention from scratch in the same rhythm.
How it teaches
How AI Systems Craft teaches the subject.
The method in AI Systems Craft is to make every hidden step visible. A concept is introduced in plain language, translated into code, run with exact output, then checked with questions that ask you to predict what changed. Instead of dropping the reader into abstract formulas, the lessons build intuition from shapes, counts, comparisons, error measurements, and repeated experiments. Basic arithmetic is enough to begin: addition, subtraction, multiplication, division, percentages, and a willingness to follow a small program line by line.
Assumed background
No advanced math is assumed. You should be comfortable with basic arithmetic and willing to install or use Python, but the book explains the needed data, metric, and model ideas as they appear.
Reasonable timeframe
A realistic self-study pace is 8 to 12 weeks if you work through several lessons per week and actually run the projects. A faster reader can skim the early data chapters, but beginners should treat the book as a lab manual: read, run, inspect, and repeat. By the end, the reader should be able to reason about data preparation, classical models, neural-network basics, embeddings, RAG, evaluation, cost, scaling, and MLOps as connected engineering decisions.
What a study session looks like.
01
The idea behind it
Each chapter opens with why its concept matters, before any code.
02
Project and output
A complete, runnable program with the exact output to check your prediction against.
03
Logic Lens
The project read again line by line, as a sequence of small decisions.
04
Common mistakes
The traps most learners fall into, and how to sidestep them.
05
Try it
Small changes that turn the example into your own experiment.
06
Quiz and answer key
Questions at the end of every chapter, with full answers in the chapter's answer key.
Secure checkout on Stripe; no account and no subscription
A private download link on your receipt page and by email
Each link works once, so save the PDF when it downloads
Sold separately from the Mac apps
Questions about AI Systems Craft.
Is AI Systems Craft suitable for beginners?
Yes. No advanced math is assumed. You should be comfortable with basic arithmetic and willing to install or use Python, but the book explains the needed data, metric, and model ideas as they appear. It is written for self-study and builds up step by step.
How long does it take to work through AI Systems Craft?
A realistic self-study pace is 8 to 12 weeks if you work through several lessons per week and actually run the projects. A faster reader can skim the early data chapters, but beginners should treat the book as a lab manual: read, run, inspect, and repeat. By the end, the reader should be able to reason about data preparation, classical models, neural-network basics, embeddings, RAG, evaluation, cost, scaling, and MLOps as connected engineering decisions.
What format is AI Systems Craft, and how do I get it?
AI Systems Craft is an 854-page PDF. Checkout is handled by Stripe. When payment completes, the receipt page gives you a private download link, and the same link is sent to the email address you used at checkout.
Can I read it on any device, and is it mine to keep?
Yes. It is a standard PDF you can read on a computer, tablet or phone, online or offline. It is a one-time purchase for $25 with no subscription, and the file is yours to keep.
What if my download link stops working?
Each link works once and shows when it expires, so save the PDF when the download finishes. If a download fails or a link has expired, email support@macossoftware.com with your Stripe receipt and we will help.
