What a data scientist actually is

The phrase data scientist describes a family of related jobs, not a single one. Understanding the variants early helps you target your learning and, later, your job search. The clearest way to think about it is by what a person is accountable for.

1.1 The three tracks#

1. Product and analytics data scientist. Uses SQL, metrics, experimentation, and product sense to drive decisions. At companies like Meta this role sits close to the product team. If you like framing questions and running experiments, this is your track.
2. Machine learning and modeling data scientist. Builds models: feature engineering, training, evaluation, and often some deployment. If you like building systems that predict, this is your track.
3. Applied or research scientist. Invents or adapts methods, usually with an advanced degree, at research-heavy organizations. If you like novel problems and publishing, this is your track.

1.2 Data scientist versus neighboring roles#

1.3 The modern data science workflow#

Whatever the track, the work follows a recognizable loop. Most beginners imagine data science is mostly modeling. In reality, framing the problem and preparing the data take the majority of the time, and communicating the result is what makes the work matter.

1.4 The skills this guide builds#