Data science asks for two different kinds of skill. There is the technical half: programming (Python is usually enough), statistics, probability and machine learning. Then there is the half that decides whether any of it gets used — leading a team, making a case with data, and explaining a result to someone who will act on it. I would note that even if you aren't a manager, you should understand management.
Here is what helped me, hard skills first.
Statistical Foundations
1. All of Statistics by Larry Wasserman
Statistical theory in one short, rigorous volume. It is the book that fills in the maths under everything else on this list.
2. The Elements of Statistical Learning by Trevor Hastie, Robert Tibshirani, and Jerome Friedman
The reference for statistical learning methods. Dense — I use it to look things up rather than to read straight through.
3. 100 Statistical Tests by Gopal K. Kanji
A lookup table for picking the right test. Where I go when designing an experiment or reading an A/B test.
4. The Book of Why: The New Science of Cause and Effect by Judea Pearl and Dana Mackenzie
Pearl gives you a way to say what causes what, rather than what moves with what. It changed how I think about experiments. For the technical treatment, see his earlier Causality. Also worth reading: Data Analysis Using Regression and Multilevel/Hierarchical Models by Gelman and Hill.
Machine Learning & Big Data
5. Mining of Massive Datasets by Jure Leskovec, Anand Rajaraman, and Jeff Ullman
Algorithms and systems for data that does not fit on one machine. Free online.
6. Machine Learning Course by Andrew Ng
Still the clearest introduction I know of, and the exercises are the part that teaches.
7. Linear Algebra Course by Gilbert Strang (MIT OpenCourseWare)
Linear algebra is the language machine learning is written in. Strang shows you what the operations are doing, not just how to do them.
Leadership & Team Building
8. The Carrot Principle by Adrian Gostick and Chester Elton
Recognition, treated as a tool with a method rather than a nice gesture.
9. High Output Management by Andrew Grove
Grove's argument is that a manager's output is the output of their team, so the job is finding the actions with the most leverage. That transfers directly to running a data science team.
Communication & Influence
10. Influence: The Psychology of Persuasion by Robert Cialdini
An analysis is worth nothing until someone acts on it. Cialdini is about how a recommendation gets framed so that happens.
11. The Visual Display of Quantitative Information by Edward Tufte
This is a great coffee table book that gets you thinking about visualization. I would note though that often the most impactful visualizations are quite simple and accompanied with clear calls to action.
12. How to Know a Person by David Brooks
Brooks is on listening, and on what it takes to actually understand the person across from you. Most of leading people runs through that.
Innovation & Technology
13. The Innovator's Dilemma by Clayton Christensen
Why good companies lose to worse products, and how to place a data science bet inside one.
14. The Soul of a New Machine by Tracy Kidder
Kidder followed a team building a computer in the late 1970s. It won a Pulitzer, and it is the best description I have read of what engineering work actually feels like.
Critical Thinking
15. Justice Course by Michael Sandel (Harvard)
The systems we build make decisions about people. Sandel is where I go for how to think about the ones that are not technical questions at all.