“What knowledge do I need for a career in machine learning (ML)? How do we transform our data science projects into production systems? Do you have examples of real production ML use-cases?” ML enthusiasts, from scientists and engineers up to business leaders, often ask me those questions.

For individuals and organizations alike, the road to impactful ML is exciting yet bewildering: tools change fast, the scientific bar intimidates, the community boils with new ideas and the outer world is confused. Yet, when you start working on real ML use-cases, you realize that while tools are changing, relevant algorithms are not changing as fast. While the equations behind some algorithms are daunting, many of them use teenager-level math. While there are blog posts and research papers published every day, you only need to read a fraction of them – the right fraction - to delight your customers and yourself. Becoming an effective ML practitioner does not require super-human reading abilities nor a Fields Medal. You can be impactful and happy by investing in targeted knowledge gaps, learning skills that matter and helping others. All you need is to identify those areas deserving attention on your ML journey.

In this book, I present the differentiators that make ML impactful. I share learnings and best practices that help embrace a career in ML and that successful ML organizations and practitioners apply.

Target audience

In this book I answer three questions:

1. What are AI and ML in practice?
2. What baseline knowledge is needed to embrace a technical career path in ML?
3. Which skills are rare and valuable for ML practitioners and organizations?

This book is valuable for multiple personas:

• Product managers and business leaders will learn ML success factors and discover associated best practices and field stories.
• Individuals considering a career in ML will be able to prioritize their learning path and accelerate their transition.
• Experienced ML practitioners will learn new techniques, use-cases and stories to sharpen their capabilities and increase their impact.

What this book is not

There are dozens of great books about machine learning. My goal is to complement them and not overlap them. This book focuses on topics not extensively covered yet in ML literature, and deliberately omits few topics.

• This book is not an ML coding book. It contains almost no code. If you are interested in learning ML code, I recommend the following books: Python Data Science Handbook by Jake VanderPlas, Deep Learning with Python by François Chollet, Deep Learning for Coders by Jeremy Howard and Sylvain Gugger or Dive into Deep Learning by Zhang et al.
• This book is not an ML science book. It contains almost no equation. If you are interested in ML science, I recommend the following books: The Elements of Statistical Learning: Data Mining, Inference, and Prediction by Hastie et al. and Pattern Recognition and Machine Learning by Bishop. Deep learning and its practical implementation are well covered in Dive into Deep Learning by Zhang et al.
• This book does not cover all ML use-cases. I focus on approaches mature enough to consider them part of baseline knowledge for production. For example, although interesting and promising, I do not cover reinforcement learning nor generative adversarial networks (GANs).
• This book does not cover ML interpretability, already well covered in Christoph Molnar’s book, Interpretable Machine Learning. As AI regulation and compliance frameworks mature, they will influence the convergence and refinement of model inspection tooling, which is still in its infancy and changing fast.