After a lot of writing, rewriting, editing, diagram-fixing, code-checking and the inevitable late-stage “why did I phrase it like that?” moments, I’ve finally completed the proofs for the second edition of Advanced Data Science and Analytics with Python.
The new edition reflects just how dramatically the AI landscape has shifted since the first edition appeared in 2020. What started as an update quickly became a substantial expansion of the book itself. I love the fact that the copy editor mentions that it’s nice to work together again (three times already – see screenshot).
This edition now includes significantly broader coverage of deep learning, reinforcement learning, generative adversarial networks, transformer architectures, vector search, graph representation learning and modern NLP workflows. Generative AI takes a much more central role, with dedicated sections on self-attention, BERT, GPT models, large language model evaluation and API-based interaction. There is also new material on emerging agentic systems and on-device Foundation Models.
One of the biggest challenges was resisting the temptation to turn the book into a 4,000-page monument to every new AI paper announced before breakfast. The field is moving absurdly quickly, so the real goal was to focus on durable concepts, practical implementation and architectural understanding rather than hype-cycle archaeology.
The Python ecosystem has evolved enormously as well, so a large amount of the code, tooling and deployment discussion has been modernised throughout. The aim remains the same: bridge the gap between theory, implementation and production-grade thinking.
Now comes the strange post-proof phase where, after staring at the manuscript for a year, I will almost certainly open the printed copy to discover a typo within 14 seconds.
More details soon.
