What does BLS say about education?
The Bureau of Labor Statistics Occupational Outlook Handbook gives a bachelor's degree as the typical entry-level education for data scientists. It lists no required prior work experience and no on-the-job training in its quick facts. It also says some employers require or prefer a master's or doctoral degree. The common degree fields it names are mathematics, statistics, computer science, business and engineering.
BLS stresses math. It points high school students toward linear algebra, calculus, and probability and statistics, and says college students need computer science on top of math and statistics. O*NET puts the occupation in Job Zone Four, where most jobs need a four-year bachelor's degree but some do not. So a degree is the usual route. It is the norm, and O*NET leaves room for exceptions.
Which skills should you build first?
Start with statistics, then Python and SQL. O*NET publishes Lightcast data on US job postings from 2025 that are linked to the data scientist occupation. Python appeared in 66% of those postings, SQL in 51%, and R in 34%. Tableau and Power BI followed at 22% and 19%. Cloud skills showed up too, with AWS at 17% and Azure at 13%. TensorFlow and PyTorch each appeared in roughly one posting in ten.
Read those numbers as a priority list. Python and SQL are close to table stakes. R is worth learning if you come from statistics or research. A visualization tool matters because BLS says data scientists present findings to nontechnical audiences and make business recommendations. Deep learning frameworks matter most if you lean toward machine learning work.
How analyst roles can build experience
O*NET says jobs in this zone usually need several years of work-related experience, on-the-job training or vocational training. Analyst jobs, such as data analyst or business intelligence analyst, are one way to build that experience. Those jobs teach you messy data, SQL against production tables, and how to explain a result to a manager who has ten minutes.
Domain knowledge counts as well. BLS gives the example of asset management firms that may want finance experience or coursework in investments or banking. If you already work in healthcare, finance, retail or logistics, an analyst job in that same industry is often the shortest route, because you bring context other applicants lack.
What should a portfolio show?
BLS lists the core duties: find useful data, collect and clean it, build and test models, visualize results, and make recommendations. A good portfolio covers that whole loop on two or three projects. Pick raw, messy data, document how you cleaned it, compare a simple baseline against a stronger model, and write up what a business should do with the answer.
Put the code in a public repository with a readable README. A portfolio should show the work BLS and O*NET describe: sound statistics, working SQL and Python, and honest testing of your models, including where they fail. One finished project with a clear write-up beats five half-done notebooks copied from a course.
Pay
USD a year, before tax. US figures. BLS rows use OEWS May 2025 data for the data scientists series, moved to 2026 with the Employment Cost Index. That series covers all data scientists, not AI specialists alone. Job ad rows are midpoints of US dollar pay ranges in 63 live AutomationRoles.ai ads from 27 employers on 5 November 2026. They lean toward large AI and tech employers and show base pay only, without stock or bonus. Refer to the AI Data Scientists salary guide for figures from live adverts.
What you need
- Bachelor's degree (typical)BLS gives a bachelor's degree as the typical entry education, often in math, statistics, computer science, business or engineering. Some employers require or prefer a master's or doctorate.
- Statistics and mathBLS points to linear algebra, calculus, and probability and statistics as the foundation. You need them to design, validate and explain models.
- PythonPython appeared in 66% of 2025 US data scientist job postings in the Lightcast data O*NET publishes. Treat it as your main language.
- SQLSQL appeared in 51% of those postings. You will pull and shape most of your data with it.
- CommunicationBLS says data scientists must explain results to technical and nontechnical audiences and turn analysis into business recommendations.
- Industry knowledgeSome employers want industry experience or coursework, such as finance for asset management roles, according to BLS.
- No required credentialNo regulator oversees data science work and no certification is required to work in it. Certifications can help show cloud or tool skills, but they do not replace a portfolio.
How to get there
- Years 1-4Earn a quantitative bachelor's degreeMath, statistics, computer science, engineering or business with heavy stats coursework. Take linear algebra, calculus and probability.
- Years 2-4Learn Python, SQL and core statisticsBuild these alongside your degree. Add R or a visualization tool such as Tableau or Power BI once the basics feel routine.
- Final year and afterBuild two or three portfolio projectsUse messy real data, document cleaning, compare models, and write a plain-language recommendation for each.
- Years 1-3 of workStart in an analyst roleData analyst, BI analyst or analytics engineer jobs give you the work experience O*NET says most of these jobs need.
- After several yearsMove into a data scientist titleTake on modeling and experiment work in your analyst job, then apply internally or externally with your portfolio.
- Any point, if neededConsider a graduate degreeOptional. Useful if target employers require or prefer a master's or doctorate, which BLS says some do.
Open now on AutomationRoles.ai
Questions people ask
Do you need a master's degree to become a data scientist?
Not usually. BLS gives a bachelor's degree as the typical entry education, though it notes some employers require or prefer a master's or doctoral degree.
Can you become a data scientist without a degree?
It is harder. O*NET says most jobs in this group need a bachelor's degree but some do not. Without a degree, you need strong work experience, usually from analyst roles, and a portfolio that proves your statistics and coding.
Should I learn Python or R first?
Python first for most people. In the 2025 Lightcast posting data O*NET publishes, Python appeared in 66% of US data scientist postings and R in 34%. Learn SQL right after, since it appeared in 51%.
How long does it take to become a data scientist?
Plan on a four-year degree plus several years of related work. O*NET says jobs in this zone usually need several years of work-related experience or training. Career changers who already have a quantitative degree can move faster.
Is demand for data scientists growing?
BLS projects employment of data scientists to grow 35% from 2025 to 2035, much faster than average. It expects about 24,800 openings a year on average, many to replace people who leave the field.
How much do data scientists earn in the US?
BLS reports a median of $120,230 in May 2025. Moved to 2026 terms, the median is about $124,000, with the middle half between about $88,300 and $163,900. Pay ranges in ads on this board run higher. The ads lean toward large AI and tech employers, show base pay only without stock or bonus, and are not a market average.
