Career Guide
Data Scientist Career Guide
What data scientists do all day, where they work, how people get into the job and what the degree costs. Pay and outlook come from the Bureau of Labor Statistics, the duties from O*NET, and the education costs from our own rankings.
- Median annual pay
- $120,230
- BLS OEWS, May 2025
- Projected growth, 2025 to 2035
- 34.6%
- BLS Employment Projections 2025-35
- Openings a year
- 24,800
- BLS Employment Projections 2025-35
- Typical entry education
- Bachelor’s
- BLS Employment Projections
On this page
What Does a Data Scientist Do?
Data scientists build statistical and machine learning models that predict an outcome, sort records into groups or explain what drives a result. A typical project starts with framing the question, moves through collecting and cleaning data and testing several models, and ends with putting the best one into use and checking that it keeps working.
Compared with an analyst, a data scientist spends more time writing code, usually Python, and more time on modeling and experiment design. The job sits close to engineering at some companies and close to research at others, so two people with the same title can have quite different days.
A data scientist job description usually covers the whole life of a model. It starts with framing: turning a loose goal, such as keeping more customers, into a question a model can answer, and deciding how success will be measured. Then comes the data work, pulling records from warehouses and logs, cleaning them and building the features a model learns from. The modeling stage means trying several methods, from a regression to gradient boosting or a neural network, and testing each on data it has not seen. The last stage is putting the model to use, explaining what it does to the people who rely on it, and checking that it keeps working as the data changes.
How much of each stage a posting covers depends on the team. In a product or research group the job leans toward experiments and statistics; in a platform or engineering group it leans toward code, pipelines and deployment, and the title shades into machine learning engineer. The duties list tells you which version a posting describes.
BLS counts the job as data scientists, a group of 262,440 workers in May 2025. O*NET’s task list, quoted below, is the most detailed public account of the work.
Data Scientist Job Description: The Core Tasks
The data scientist job description in O*NET, the Labor Department’s occupation database, ranks task statements by how important workers say they are. These are the first four.
- Analyze, manipulate, or process large sets of data using statistical software.
- Create graphs, charts, or other visualizations to convey the results of data analysis using specialized software.
- Test, validate, and reformulate models to ensure accurate prediction of outcomes of interest.
- Deliver oral or written presentations of the results of mathematical modeling and data analysis to management or other end users.
Quoted from O*NET OnLine, Data Scientists, in O*NET’s order of importance.
Data Scientist Skills and Tools
The data scientist skills employers ask for come down to a short list of tools plus the statistics and judgment to use them. O*NET counts how often each tool appears in job postings for the occupation. In 2025, the tools named most often were Python (66% of postings), SQL (51%) and R (34%). Python alone shows up in 66% of data scientist postings, so that is the one to learn first.
Python, R and statistics software
In postings: Python 66% R 34% SAS 9% MATLAB 5%
In O*NET’s task list for data scientists, this is the part of the job it serves: “Analyze, manipulate, or process large sets of data using statistical software.” It also lists “Create graphs, charts, or other visualizations to convey the results of data analysis using specialized software.”
SQL and databases
In postings: SQL 51% Snowflake 8% NoSQL 3% Redshift 2%
SQL appears in 51% of data scientist postings, ahead of Snowflake at 8%; O*NET’s task list for data scientists doesn’t name this kind of tool, so treat it as a skill employers screen for rather than the core of the data scientist job.
Dashboards and BI tools
In postings: Tableau 22% Power BI 19% Looker 3% Alteryx 2%
In O*NET’s task list for data scientists, this is the part of the job it serves: “Deliver oral or written presentations of the results of mathematical modeling and data analysis to management or other end users.”
Spreadsheets
In postings: Excel 8% Access 2%
Excel appears in 8% of data scientist postings, ahead of Access at 2%; O*NET’s task list for data scientists doesn’t name this kind of tool, so treat it as a skill employers screen for rather than the core of the data scientist job.
Share of job postings naming each tool
Outside these groups, postings for the occupation also often name AWS (17%), Azure (13%), TensorFlow (11%) and PyTorch (10%). Those depend more on the employer’s own systems and are usually learned where they are used.
What Employers Expect
Hiring managers look for evidence on three fronts: that you understand the statistics behind the methods, that you can write code a team can maintain, and that you can tell whether a model is actually working. Interviews often include a coding exercise in Python or SQL, a statistics or probability discussion, and a case where you talk through how you would frame and evaluate a model for a business problem.
Experience counts in a broad sense. A research project, a thesis, an internship or modeling work in an analyst job all show that you have handled real data under real constraints. The O*NET survey below shows what workers in the occupation say a new hire needs, a useful check on postings that ask for more than the job does.
- Typical entry education
- Bachelor’s degree
- Work experience in a related occupation
- None
- Typical on-the-job training
- None
BLS Employment Projections, table 1.2: the typical entry requirements.
What workers say a new hire needs
Where Data Scientists Work
Data scientists work wherever an organization has enough data to model and decisions worth automating. The largest employers are corporate head offices, technology and consulting firms, insurers, banks and research firms, and each brings its own kind of problem: pricing and risk at an insurer, fraud and credit at a bank, demand forecasting and recommendations at a technology company.
Industry and team size
Team structure varies as much as industry. A large employer may run a central data science group with shared tools and review, or place scientists inside product teams. A smaller one may have a single data scientist who also builds the pipelines and dashboards. If you are switching careers, the industry you already know can be the easiest place to start, because knowing the business shortens the time it takes to frame a useful problem.
The largest employers of data scientists are spread across several industries, and the median differs by industry, so where you work moves pay as well as the kind of problems you solve. Among the twenty largest employing industries, media streaming distribution services, social networks, and other media networks and content providers pays the highest median, $169,080.
Data Scientists, national industry-specific estimates (4-digit NAICS), BLS OEWS May 2025. Share is the industry’s employment divided by the occupation’s national employment, 262,440.
The twenty largest employing industries
| Industry | Employed | Median |
|---|---|---|
| Management of Companies and Enterprises | 28,620 | $128,050 |
| Computer Systems Design and Related Services | 27,590 | $132,380 |
| Insurance Carriers | 15,090 | $107,680 |
| Management, Scientific, and Technical Consulting Services | 15,060 | $112,520 |
| Credit Intermediation and Related Activities | 12,450 | $130,300 |
| Scientific Research and Development Services | 11,480 | $131,420 |
| Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services | 11,320 | $129,590 |
| Software Publishers | 10,950 | $156,220 |
| Employment Services | 8,050 | $102,560 |
| Agencies, Brokerages, and Other Insurance Related Activities | 7,420 | $110,880 |
| Securities, Commodity Contracts, and Other Financial Investments and Related Activities | 7,210 | $134,510 |
| General Medical and Surgical Hospitals | 7,050 | $101,520 |
| Colleges, Universities, and Professional Schools | 6,220 | $81,120 |
| Other Professional, Scientific, and Technical Services | 4,610 | $103,210 |
| Nondepository Credit Intermediation | 4,050 | $126,650 |
| Architectural, Engineering, and Related Services | 4,040 | $105,400 |
| Media Streaming Distribution Services, Social Networks, and Other Media Networks and Content Providers | 3,660 | $169,080 |
| Office Administrative Services | 3,570 | $99,790 |
| Accounting, Tax Preparation, Bookkeeping, and Payroll Services | 3,320 | $128,120 |
| Professional and Commercial Equipment and Supplies Merchant Wholesalers | 2,790 | $129,210 |
For where data scientists jobs cluster by state and metro area, and how pay compares in each, see the data scientist salary report.
How to Become a Data Scientist
Most data scientists start with a degree in a quantitative field, then add the programming and machine learning the job runs on. Employers look for three things: statistics you understand well enough to choose a method and defend it, Python or R you can write cleanly enough for someone else to rerun, and finished projects that go from a raw dataset to a tested model and a written explanation of what it does.
A sensible learning order
- Statistics and SQL
- Python and its data libraries
- Machine learning methods
- Evaluating models honestly
- Cloud tools and version control
Statistics comes first because it decides which method fits a problem and how to judge the result; cloud tools and version control are usually learned along the way. Many people get there in stages: a first job as a data analyst or software engineer, then a move into data science once their work starts to include modeling. A master’s in data science or statistics is a common way to make that move with structure.
BLS lists a bachelor’s degree as the typical entry education. These are the common routes in. Degree lengths are NCES’s definitions; credits are the median of the programs in our rankings that state a credit total.
-
A bachelor’s in a related fieldusually at least four years of full-time study
The usual starting point. Common majors are statistics, computer science, mathematics, physics, economics or engineering. BLS lists a bachelor’s degree as the typical entry education. Pair the degree with Python and SQL, the tools data scientist postings name most, so a hiring manager can see applied work.
See every master’s ranking → -
An online master’s in data sciencegenerally one or two years of full-time study beyond the bachelor’s · median 30 credits
For a bachelor’s holder who wants the depth the data scientist job leans on; the first task O*NET lists for it is to analyze, manipulate, or process large sets of data using statistical software. The 25 programs on our Data Science list run from $9,900 to $61,712 in total tuition, and most are built to fit around a full-time job.
Compare data science programs → -
A master’s from our Best Overall listgenerally one or two years of full-time study beyond the bachelor’s · median 31 credits
Fully online data analytics master’s programs, business school degrees included, scored on cost, graduation rate, track record and flexibility. Its 25 programs have a median total tuition of $22,050, a narrower fit for data scientists who already know which part of the data scientist work they want.
Compare best overall programs → -
A related degree plus experienceusually at least four years of full-time study, then work experience
Many data scientists arrive from data analysts and statisticians roles, building skills on the job and with short courses. A portfolio of finished projects in Python carries most of the weight on this route.
Compare master’s programs without a GRE requirement →
Programs to Start With
Programs from the rankings that fit this career, led by those whose own pages point toward data scientist work, with the score and rank each carries there.
Can You Become a Data Scientist Without a Degree?
BLS lists a bachelor’s degree as the typical entry education.
It is uncommon. The job rests on statistics that are hard to show without coursework, and the O*NET survey in the skills section above shows how many workers in the occupation say a new hire needs a graduate degree. People who get in without one usually come from software engineering or another technical job, with published code or project work that shows they can build and test models.
A more reliable route is a degree in a quantitative field, then an analyst job, then a move into modeling work. If you already hold a bachelor’s in another field, an online master’s can supply the statistics and machine learning while you keep working.
Entry-Level Data Scientist Jobs
Entry-level data science jobs carry titles such as junior data scientist, associate data scientist or data scientist I, and some first roles are listed as data analyst or analytics engineer even when the work includes modeling. Internships and research assistant roles are a common way in for students, and many employers fill junior roles from their own analyst teams.
The first year is usually spent on scoped problems with review from senior staff: improving an existing model, writing up an experiment, or building a baseline someone else will extend. People who do well early keep careful records of each experiment, compare every model with a simple baseline, and learn how the company deploys and monitors models, not only how to train them.
Titles to search for: Junior Data Scientist · Data Scientist I · Associate Data Scientist
The best federal guide to starting pay is each occupation’s 10th percentile, the pay below which one in ten workers fall. For data scientists that is $67,240, and the 25th percentile is $85,660. Many first jobs sit in the lower part of the range, but that is an inference, not a published figure. The data scientist salary report shows every percentile.
What the Degree Costs
Total tuition for the fully online master’s programs in our rankings, each read from the school’s own tuition page and verified.
In data science a master’s is common enough that its price deserves a close comparison. The figures above are total tuition for fully online programs we verified on each school’s own page, so they can be set side by side in a way per-credit rates cannot. Programs differ widely in price, and the price says little about how much machine learning, programming or statistics the course list holds.
Three questions narrow the choice. Does the curriculum go beyond tools into statistical learning and model evaluation, with a capstone that ends in a working model? Can you keep up with it while working? And will an employer pay for part of it? Someone already working as an analyst may find an employer willing to share the cost of a move into modeling work.
These totals are tuition only, before fees, books and any employer or financial aid. A lower per-credit rate on a longer program can cost more overall, so compare totals. For what the career pays in return, the data scientist salary report shows pay in every state, the top-paying metros and industries.
Data Scientist Salary and Job Outlook
Data scientists have their own federal occupation code, so the pay and outlook figures below describe the job directly rather than a set of neighbors. BLS counts workers by what they do, not by title, so the code covers anyone whose work fits the occupation’s definition, and O*NET files business intelligence analysts under it as a more specific profile.
The spread is wide. BLS publishes percentiles rather than pay by experience, so the range below shows how far apart pay runs within one occupation, not a salary ladder. Industry, location and the scope of the role all move where a given job lands, and the highest-paying state, metro and industry show how far. On outlook, BLS ties its projection to the growing amount of digital data organizations collect, which keeps demand for people who can model it rising across industries.
- 10th percentile
- $67,240
- 25th percentile
- $85,660
- Median
- $120,230
- 75th percentile
- $158,880
- 90th percentile
- $199,130
Half of data scientists earned more than $120,230 in May 2025 and half earned less; the middle half fell between $85,660 and $158,880. BLS reports pay as percentiles across the occupation, so this guide gives the range rather than a ladder of salaries. For what moves data scientist pay by state, metro area and industry, and how far it stretches after local prices, see the data scientist salary report.
BLS’s stated reason for the projection: “Demand change - share increases as larger amounts of digital and electronic data are collected over the projections decade. Businesses in all industries will hire data scientists to analyze data to help improve business processes and design and develop new products.”
Data Scientist vs. Data Analyst
The two jobs start from the same data and often sit on the same team, so the line between them moves from company to company. The clearest split is the question each one answers. A data analyst explains what happened and why, with queries, summaries and dashboards. A data scientist builds models that predict what will happen or make a decision automatically, such as which transactions look like fraud or which customers are likely to cancel.
The difference shows up in what employers ask for: data scientist postings lean on Python and machine learning libraries, and analyst postings lean on SQL and dashboard tools. Many data scientists started as analysts, and the move usually comes from adding statistics and programming depth through work projects, a master’s or both.
O*NET’s posting counts put numbers on the split. Python appears in 20% of postings for business intelligence analysts (the closest O*NET profile for data analysts) and 66% for data scientists; SQL appears in 35% and 51%.
Data Scientist vs. Data Analyst at a glance
| Data Scientist | Data Analyst | |
|---|---|---|
| The question it answers | What will happen, or what to do automatically | What happened, and why |
| Main output | Tested models and experiment results | Queries, dashboards and written findings |
| Pay (BLS median) | $120,230 | $78,760 to $120,230 across related jobs |
| Tools named most in postings | Python 66% · SQL 51% · R 34% | SQL 35% · Python 20% · Power BI 20% |
Read the full data analyst career guide for its pay, tasks and degree options.
Data Scientist Career Path: How the Career Moves
A data science career widens in scope more than it changes in kind. Early on the job is building models for problems someone else framed; later it is choosing which problems deserve a model at all, and knowing when a simpler method or a plain analysis will do. Promotion tends to follow impact: a model that is used, trusted and kept working counts for more than a long list of experiments.
Typical stages. Titles and scope change by employer; BLS reports one pay range for the whole occupation.
-
Entry-level data scientist
Junior Data Scientist · Data Scientist I · Associate Data Scientist
Build and evaluate models on scoped problems with review from senior staff. Learning the data, the deployment process and how results are judged takes most of the first year.
-
Data scientist
Data Scientist · Machine Learning Scientist
Own projects from framing to deployment, choose methods and explain trade-offs to stakeholders. You are trusted to decide which approach fits, not just to run it.
-
Senior data scientist
Senior Data Scientist · Lead Data Scientist
Lead projects that span teams, set modeling standards and mentor others. Knowing when a simpler model is good enough becomes as important as building complex ones.
-
Principal or manager
Principal Data Scientist · Data Science Manager · Director of Data Science
The path splits. Principals set technical direction across many projects; managers hire, develop and organize the team and its roadmap.
Where the path splits
At the senior stage the path splits. Some people stay technical and move toward principal roles, machine learning engineering or research, setting methods and standards across teams. Others move into management, where the work becomes hiring, setting priorities and making sure the team’s models reach the people who need them. Titles vary between employers, so compare the scope of a role, not its label.
Is Data Scientist a Good Career? Pros and Cons
What draws people to the work
The appeal is the mix. The job combines statistics, programming and real business problems, a model that works can shape a large number of decisions every day, and the skills carry across industries. BLS projects the occupation to keep adding jobs, as the figures above show.
- The work is technical and varied, mixing statistics, programming and problem framing.
- Strong demand across tech, finance, healthcare and government.
- Models you build can run at a scale that affects thousands of decisions a day.
- Clear paths into machine learning engineering, research and leadership.
- Projected growth of 34.6% for 2025 to 2035, against 3.5% for all occupations.
What to weigh first
The downsides deserve the same attention. Many projects never reach production, the bar for statistics and programming is high and keeps moving as tools change, and leadership can expect more from the data than it can deliver. The way in also tends to be longer than for an analyst job: in the O*NET survey above, a large share of workers say a new hire needs a master’s.
- Many projects never reach production, which can be frustrating after weeks of work.
- The bar for statistics and programming is high, and the tools change quickly.
- Expectations from leadership can run ahead of what the data can support.
- A wide pay range: a tenth of data scientists earn under $67,240, so the median isn’t a typical offer.
Worth knowing
Which side weighs more depends on the employer. A team with clean data, a working deployment process and managers who understand uncertainty removes most of the frustrations above, so ask about all three in an interview before you accept an offer.
Is It for You?
The questions below are about temperament as much as skill. Data science suits people who like statistics and programming about equally, who can sit with an experiment that fails and learn from it, and who care whether a model gets used once it is built.
Five questions before you commit. If most of these get a yes, the job is likely a good match.
- Do you enjoy statistics and programming in roughly equal measure?The job needs both every day, and weakness in either shows quickly.
- Can you live with most experiments failing?Many models never beat the simple baseline, and that is a normal result.
- Can you explain uncertainty to people who want a yes or no?Stakeholders often want certainty that the data can’t give.
- Will you keep up with a field that changes fast?Methods and libraries move quickly, and staying current is part of the work.
- Do you care whether your work gets used?Getting a model into production and trusted takes as much effort as building it.
If several get a no, look at the neighbors. Someone who prefers answering questions to building models may be happier as a data analyst, someone drawn to the mathematics may prefer statistics, and someone who enjoys production systems most may find machine learning engineering a closer match. The related careers below cover the first two.
Skills can be learned along the way, and nobody starts with all of them. The questions point to the parts of the job that wear on people who chose it for the title or the pay, so an honest answer now saves a harder change later.
A Day in the Life of a Data Scientist
A data scientist’s day moves between three kinds of work: keeping existing models healthy, framing new problems with the people who will use the results, and the longer stretches of experiments and code. The schedule below is a composite built from the job’s core tasks, not a record of one person’s day.
- MorningCheck how deployed models performed overnight and look into any drift in their predictions.
- Late morningMeet with a product team to frame a new problem: what to predict, how success will be measured and what data exists.
- AfternoonEngineer features and compare several models in a notebook, keeping track of what each experiment changed.
- Late afternoonReview a colleague’s analysis and talk through whether the results justify the conclusion.
- End of dayWrite up the experiment so the team can see what worked, what didn’t and what comes next.
Much of the experimenting is unglamorous: checking that the training data matches what the model will see in use, rerunning a comparison after fixing a bug, writing down which change made the difference. Teams that keep careful experiment records spend less time repeating work, and a short written summary at the end of the day makes the next morning faster.
The mix changes over a project. The first weeks go to framing and data work, the middle to experiments, and the end to review, deployment and explaining the model to the people who will rely on it, which often takes longer than expected.
Related Careers
Data science sits between statistics, software engineering and analysis, and people move in all three directions. The careers below share much of the same toolkit; what changes is the main output, from study design to dashboards to optimization models, and with it the degree employers expect.
National medians and projected growth for the careers that sit closest to this one.
$78,760 to $120,230 range
$105,650 median
$139,500 median
$120,230, data scientists median
Choosing between them
If you are deciding between them, look at what you would spend most of the week producing. A statistician spends it on study design and inference, a data analyst on reports and findings, a business intelligence analyst on reporting systems, and an operations research analyst on optimization and planning models.
Choosing between them
Moves between these roles are common in both directions. An analyst who adds modeling depth can move into data science, and a data scientist who prefers designing studies or building planning models can move toward statistics or operations research without starting over.
How it compares with related careers
| Occupation | Median | Growth, 2025-35 | Openings a year | Typical entry education |
|---|---|---|---|---|
| Database Architects | $139,500 | 9.4% | 3,900 | Bachelor’s degree |
| Data ScientistsThis page | $120,230 | 34.6% | 24,800 | Bachelor’s degree |
| Computer Systems Analysts | $105,850 | 7.9% | 32,900 | Bachelor’s degree |
| Statisticians | $105,650 | 11% | 1,900 | Master’s degree |
| Operations Research Analysts | $88,940 | 11.9% | 7,500 | Bachelor’s degree |
| Market Research Analysts and Marketing Specialists | $78,760 | 7% | 82,000 | Bachelor’s degree |
Start with the degree
Compare fully online master’s programs on total tuition, graduation rate, size and flexibility.
Data Scientist Career FAQs
What does a data scientist do?
Data scientists build statistical and machine learning models that predict an outcome, sort records into groups or explain what drives a result. A typical project starts with framing the question, moves through collecting and cleaning data and testing several models, and ends with putting the best one into use and checking that it keeps working. O*NET’s top task for the occupation is to “analyze, manipulate, or process large sets of data using statistical software.”
How much does a data scientist make?
BLS puts the national median for data scientists at $120,230 a year (May 2025). The 10th percentile was $67,240 and the 90th was $199,130, so four in five workers earn somewhere in that span. The salary report breaks pay down by state, metro area and industry.
Is data scientist a good career?
On BLS’s figures it pays well above the typical job, with a median of $120,230, and employment is projected to grow 34.6% for 2025 to 2035 against 3.5% for all occupations. Whether it suits you depends on the questions in the fit check above.
What is the difference between a data scientist and a data analyst?
A data analyst explains what happened and why, mostly with SQL, spreadsheets and dashboards. A data scientist builds models that predict outcomes or make decisions automatically, and leans more on Python, machine learning and experiment design. The jobs overlap and often share a team, and many data scientists started out as analysts before adding statistics and programming depth.
Can a data analyst become a data scientist?
Yes, and it is a common path. The move usually comes from adding depth in statistics, Python and machine learning, then taking on modeling work in your current job so there is real experience to point to. A master’s in data science or statistics is one structured way to make the switch, and many online programs can be finished while working.
Related Pages
Sources
Every figure on this page comes from these sources. The access date on each one shows when we last pulled it.
- Occupational Employment and Wage Statistics, May 2025 (opens in a new tab)U.S. Bureau of Labor Statistics · Pay by occupation, state, metro and industry · Accessed Nov 17, 2026
- Employment Projections 2025-35 (opens in a new tab)U.S. Bureau of Labor Statistics · Projected employment change and openings · Accessed Nov 17, 2026
- Employment Projections, table 1.2 (opens in a new tab)U.S. Bureau of Labor Statistics · Entry education, experience and training · Accessed Nov 17, 2026
- Data Scientists (15-2051.00) (opens in a new tab)O*NET OnLine · Tasks, titles, software and education survey · Accessed Nov 17, 2026
- Hot Technologies for Data Scientists (15-2051.00) (opens in a new tab)O*NET OnLine · Share of job postings naming each tool, Lightcast postings January 1, 2025 to December 31, 2025 · Accessed Nov 17, 2026
- Hot Technologies for Business Intelligence Analysts (15-2051.01) (opens in a new tab)O*NET OnLine · Share of job postings naming each tool · Accessed Nov 17, 2026
- Digest of Education Statistics 2022, Appendix B (opens in a new tab)NCES and the U.S. Department of Education · Standard lengths of bachelor’s and master’s degrees
- Program and tuition pages of the schools in our rankingsTuition, format, length and admissions, read on each school’s own site · Accessed November 2026