Books on Data

Tech books age like milk in a server room. That’s the usual complaint, and it’s mostly wrong. The right ones turn into your grandmother’s carburetor manual: still correct, still useful, long after the thing it describes went out of fashion.

These are the ones that survived my shelf, somewhere between “wait, what’s a JOIN” and “let me index that for you.”

Databases and SQL

T-SQL Fundamentals by Itzik Ben-Gan. The closest thing to a SQL therapist. It will gently ask whether you really need that loop.

SQL Server Query Performance Tuning by Grant Fritchey. Turns you into a reader of execution plans, which is a different skill from writing queries and much rarer. Side effect: you will want to index everything for about a month. Resist.

Systems and engineering

Designing Data-Intensive Applications by Martin Kleppmann. If you read one book on this list, this is it. It explains why distributed systems break, not just how to configure them, and almost nothing in it has expired.

Streaming Systems by Akidau, Chernyak, and Lax. Watermarks, windowing, and why “real-time” means six different things depending on who’s asking.

Data Pipelines Pocket Reference by James Densmore. Short, practical, no padding. Good for handing to someone new on the team.

Data Management at Scale by Piethein Strengholt. The full cloud lifecycle, ingestion through consumption, with enough architecture to argue with.

Data Mesh by Zhamak Dehghani. Worth reading even if you end up disagreeing, and plenty of people do. The organizational argument outlasts the implementation advice.

Data Governance by John Ladley. Unglamorous and necessary, which is also a fair description of governance work.

Analysis and machine learning

Python for Data Analysis by Wes McKinney. Written by the person who built pandas, which is the correct credential.

Python Data Science Handbook by Jake VanderPlas. NumPy, pandas, Matplotlib, scikit-learn, in the order you actually need them.

The Elements of Statistical Learning by Hastie, Tibshirani, and Friedman. Math-heavy and not a weekend read. Also the reason you’ll understand what a model is doing instead of what its API returns.

Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Aurélien Géron. The applied counterweight to ESL.

The Hundred-Page Machine Learning Book by Andriy Burkov. Does what the title says.

Deep Learning by Goodfellow, Bengio, and Courville. Dense, foundational, and the place to go when a blog post hand-waves.

Artificial Intelligence: A Modern Approach by Russell and Norvig. Predates the current wave by decades and is better for it. Search, logic, planning, all the machinery people rediscover every few years.

Business and communication

Data Science for Business by Provost and Fawcett. Bridges the gap between what a model does and why anyone should fund it.

Storytelling with Data by Cole Nussbaumer Knaflic. The one I’d hand an analyst first, because the bottleneck is almost never the query.

Competing on Analytics by Davenport and Harris. Older, strategic, still the clearest articulation of analytics as a competitive position rather than a function.

Data Analytics Made Accessible by Anil Maheshwari. A reasonable first book for someone who isn’t technical and needs to be conversant.

---

A note from 2026

This list was put together before agentic systems became a daily part of the work, and it shows. There’s nothing here on retrieval, evaluation, or building systems where a model sits between a user and a warehouse.

What’s interesting is how little that changes the list. Kleppmann still explains why your retrieval layer will fail under load. Ben-Gan still explains why the query your agent generated is slow. Russell and Norvig were writing about planning and search long before anyone shipped a product on it.

Which is roughly the point I was making when I first wrote this. The database’s soul hasn’t changed, it’s just wearing cloud-native makeup. Add “and an LLM wrapper” to that sentence and it holds up fine.