Photo: School of Data Science at the University of Virginia (CC BY 4.0)
Job description
The work ranges from cleaning and exploring raw data to building and testing machine learning models, usually writing code in Python or R to process information and run analyses. Once a model or finding is ready, data scientists visualize and present results for non-technical audiences, then collaborate with engineers to actually deploy models into production and keep an eye on how well they hold up over time. A lot of analysis leads nowhere, so the role rewards patience with dead ends as much as technical skill; demand for the role has been growing fast, and a Bachelor's degree is the typical entry requirement.
Pros
- Much faster than average growth (33.5%), among the fastest-growing roles available
- High pay ceiling near $199k with mostly independent, self-directed daily work
- Large national demand pool (23,400 jobs) as more industries lean on data
Cons
- Above-average hours, especially around model deployments or product launches
- A lot of exploratory analysis doesn't pan out, requiring tolerance for dead ends
- Models require ongoing monitoring after deployment, since performance degrades over time
A typical day
The churn model is misbehaving, and the morning is spent finding out why: feature by feature, the data scientist discovers a data leak — a field that secretly encodes the answer — which explains the too-good validation score. Rebuilding honestly drops accuracy and increases trust, the trade the discipline is built on. Midday brings the messier, truer work: an executive wants to know whether a price change caused last month's dip, and the answer requires careful causal thinking, not just a correlation and a chart. The afternoon presentation distills three weeks of modeling into five slides for people who will never read a notebook — where "the model is 84% confident" must land without overpromising. Between meetings: Python, always Python, and the eternal data cleaning that consumes more hours than any job posting admits.
Tasks at the job
- Clean, organize, and explore large datasets to find patterns or answer business questions.
- Build and test statistical or machine learning models.
- Write code, often in Python or R, to process data and run analyses.
- Visualize findings and present results to non-technical stakeholders.
- Collaborate with engineers to deploy models into production systems.
- Validate model performance and monitor for degradation over time.
Workplace & working conditions
Work happens indoors at a computer, in an office or remote setup, with most of the day spent writing code and analyzing data rather than in meetings.
The work rewards independent, exploratory thinking and tolerance for dead ends, since a lot of analysis doesn't pan out. Deadlines exist around deliverables or product launches, but day-to-day pace is generally steady and self-directed rather than externally driven minute-to-minute.
Education necessary or common
Bachelor's degree
Common college majors
Licenses & certifications
- Cloud ML certifications (AWS/Google, optional)
- Portfolio of real projects matters more than any certificate
Career path
Related fields
Courses to take
- AP Statistics — the single most job-relevant course in high school
- AP Calculus BC — the math behind the models
- AP Computer Science A — code is the daily medium
- Linear algebra if available — machine learning runs on it
- English — communicating findings is half the value
Clubs, activities & competitions
- Math team and statistics competitions
- Coding club — push for ML projects
- Science fair with a real data project
- Kaggle competitions — beginner tiers welcome students
- Sports analytics for a school team
First steps you can take now
- Learn Python with a free course and analyze a dataset you personally care about
- Enter a beginner Kaggle competition this month
- Take a free intro machine learning course and build one model end to end
- Practice explaining one statistical idea (correlation vs causation) to a friend
