The ML Handbook

From your first DataFrame to transformers and diffusion models

A hands-on handbook that takes you from basic Python to a deep understanding of data science, machine learning, and deep learning. You learn every idea three ways: the intuition, the math, and the code. You implement algorithms from scratch in NumPy before you use the library versions, and every level ends with a capstone project that proves the skills.

Start here See the full roadmap Set up your environment

Three tiers, nine levels

• 🌱 Beginner: Data Science Foundations

  ---

  Levels 0–2. Master the Python data stack, the math underneath ML (linear algebra, calculus, probability, statistics), and the full data science workflow from SQL to A/B tests.

  • Level 0: Python for data
  • Level 1: Math foundations
  • Level 2: Data science workflow

  Beginner tier

• 🔧 Intermediate: Machine Learning

  ---

  Levels 3–5. Learn how ML really works, implement the classic algorithms from scratch, then build, tune, explain, and deploy models in production.

  • Level 3: ML fundamentals
  • Level 4: ML algorithms in depth
  • Level 5: Applied ML and MLOps

  Intermediate tier

• 🚀 Expert: Deep Learning

  ---

  Levels 6–8. Build neural networks and autograd from scratch, train deep models in PyTorch, and understand transformers, LLMs, diffusion, and RL.

  • Level 6: Neural networks from scratch
  • Level 7: Deep learning with PyTorch
  • Level 8: Modern deep learning

  Expert tier

The learning path

flowchart LR
    subgraph B["🌱 Beginner"]
        L0["Python<br/>for data"] --> L1["Math<br/>foundations"] --> L2["Data science<br/>workflow"]
    end
    subgraph I["🔧 Intermediate"]
        L3["ML<br/>fundamentals"] --> L4["ML<br/>algorithms"] --> L5["Applied ML<br/>and MLOps"]
    end
    subgraph E["🚀 Expert"]
        L6["Neural nets<br/>from scratch"] --> L7["Deep learning<br/>with PyTorch"] --> L8["Modern<br/>deep learning"]
    end
    L2 -- "tier checkpoint" --> L3
    L5 -- "tier checkpoint" --> L6

The full roadmap lists every chapter and the concepts it teaches.

How every idea is taught

• Intuition

  ---

  A plain-language explanation, an analogy, or a picture, so you know what the idea is and why it exists before any symbols appear.

• Math

  ---

  The precise formulation, with every symbol defined and results derived rather than just stated. You'll be able to read papers, not just tutorials.

• Code

  ---

  A from-scratch NumPy implementation first, then the scikit-learn or PyTorch version, checked against each other with real output.

How each chapter is organized

Part	What you get
1. Why it matters	A short real story where this knowledge saved a project, or its absence sank one.
2. Concepts	Intuition → math → code, built up step by step with diagrams and figures.
3. In practice	Complete, runnable code with real output, and the common mistakes to avoid.
4. Exercises	4–6 tasks, from easy to hard: derivations, coding, and reasoning, each with a hidden solution.
5. Check yourself	Questions to answer without notes, with hidden answers.

Each chapter closes with key takeaways, further reading, and a link to what comes next.

Who this is for

• You know some Python and want to become genuinely good at data science and ML, not just copy notebooks.
• You're a software or data engineer moving into ML, and you want to understand what the libraries are doing.
• You use ML already but feel shaky on the math, or on why your models behave the way they do.

You don't need a math degree. The math foundations level builds everything you need from high-school algebra, and every formula is checked numerically in code.

Ready?

Read how to use this handbook first. It takes ten minutes and makes the rest of the handbook far more effective.

Start here Cheat sheets