How to Become a Data Scientist

Where Data Scientists have the most money left over after rent

Median pay minus estimated federal + state + FICA taxes, minus 12 months of rent at HUD's 2-bedroom Fair Market Rent. Darker green means more money left over each year. Hover any state for the breakdown.

Education and training

Most data science positions require a master's degree or PhD in a quantitative field: statistics, mathematics, computer science, physics, economics, or a dedicated data science program. The field is more academically credentialed than general software development, a bachelor's alone is possible but puts you at a disadvantage in the hiring pool.

The core competencies: statistical modeling, machine learning algorithms, programming (Python and R primarily), SQL for database querying, data visualization, and domain expertise in the industry you're working in. Bootcamps and certificates in data science exist but carry less weight than a graduate degree for senior positions.

Licensing and certification

Data science has no licensure requirements. Voluntary credentials (AWS Machine Learning Specialty, Google Professional Data Engineer, IBM Data Science Professional Certificate) demonstrate platform competency but don't replace the academic credential + portfolio that hiring managers prioritize. A strong GitHub portfolio with well-documented projects matters more than any certificate.

What the day-to-day looks like

Data scientists clean and explore datasets (this takes more time than anyone admits, often 60-70% of a project), build predictive models, run experiments (A/B tests, causal inference), create visualizations for stakeholders, and communicate findings to non-technical decision-makers.

The gap between job descriptions and reality is wide. Many "data scientist" positions are actually analytics roles (SQL + dashboards + basic statistics) rather than machine learning roles (building models that go into production). Clarify this during interviews, the work, the tools, and the satisfaction level differ enormously.

The most impactful data scientists are those who combine technical skills with business acumen: knowing which questions to ask matters more than knowing which algorithms to apply.

Stakeholder management is the underrated skill. You can build a perfect model, but if the product manager doesn't understand why it works, it won't get deployed. If the VP doesn't trust the output, it won't influence decisions. Translating statistical results into business language, "this model predicts with 85% accuracy which customers will churn, and targeting them with retention offers could save $2M annually", is what separates data scientists who ship impact from those who build models that gather dust.

Career progression

Data analyst → data scientist → senior data scientist → staff/principal data scientist → director of data science → VP of Analytics/AI → Chief Data Officer. The IC track tops out at staff/principal levels with $250K-$400K total comp at major tech companies.

Specialization matters: ML engineering (deploying models to production), NLP, computer vision, and causal inference are high-demand subspecialties. ML engineers who can bridge the gap between research and production deployment are particularly valuable and often earn more than pure research scientists.

Salary progression

Highest paying states

Where the jobs are

Salary negotiation

Data science compensation is highly market-driven. Competing offers are the strongest lever, the field is competitive enough that employers expect candidates to have multiple options. Total comp (base + stock + bonus) negotiation follows the same patterns as software engineering. PhD holders command $15,000-$30,000 premiums over master's holders for equivalent positions.

Remote data science roles are common, and geographic arbitrage (coastal salary + low-cost-of-living location) is achievable for experienced data scientists with strong track records.

What the data doesn't tell you

The "data scientist" title has become so overloaded that it's losing meaning. BLS data for this occupation blends PhD researchers building novel ML architectures at $250K+ with analysts writing SQL queries at $75K. If you're evaluating this career, the specific role and company matter more than the occupation title. Ask every prospective employer: "What does a typical week look like for someone in this role?" and "What percentage of the work is building models vs. creating dashboards?"

Frequently asked questions