Data Scientist

01What the role is

The data scientist is the one who turns a vague business question into a sharp, answerable one and then answers it, with a model, an experiment, or sometimes the honest finding that no model is needed. The title has fractured into at least three jobs across different org charts, which is exactly why salary databases struggle with it: they average three roles into one number that describes nobody you are actually hiring.

02What they do day to day

The work splits between problem framing, model and experiment building, and communicating results in a way that changes a decision. The specialization that pays is production work: shipping LLM and generative AI systems or doing genuine causal inference adds $30,000 to $60,000 over a generalist, and regulated industries stack another premium on top.³

• Framing problems and defining success metrics before touching a model.
• Building models and running experiments, increasingly with LLMs in the toolkit.
• Causal inference and statistics where the question demands rigor.
• Translating findings into decisions stakeholders act on.

03What it pays

The median base sits near $112,590 across all employers (BLS), while stock-granting tech companies push medians higher; Glassdoor's data-science-engineer average is about $150,000.¹⁴ Mid-level scientists earn $138,000 to $175,000, senior scientists $157,000 to $194,000, and real offers span $95,000 at entry to $260,000 for a principal, with total comp crossing $330,000 at top employers.²³

The Bureau of Labor Statistics projects data-science roles growing about 34 percent, well above the average for all occupations, so the demand behind these numbers is not going anywhere.¹

04Why this role matters to data centers

Data scientists are a major consumer of data center compute: training runs, large-scale experiments, and increasingly LLM workloads all land on the same GPUs a facility was built to house. As enterprises push analytics and AI onto that infrastructure, the data scientist is who converts raw compute into an insight or a model that moves the business.

05How to hire one

Match the title to the work. Pipeline-heavy roles pay and interview like engineering; model-and-experiment roles pay like data science. Posting the wrong one is how you pay a premium for the wrong skill set and refill the seat within a year. We help you name the role correctly first, then source for it.