A data scientist title can mean three different jobs and two different budgets. Before you anchor a req to a number off a salary site, here is what the role actually pays in 2026, and where those public numbers will point you at the wrong figure. If you only need the senior band, jump to our senior data scientist salary guide.
Data Scientist Salary Guide 2026
Last updated: June 15, 2026 | By Robert Ardell
In 2026, U.S. data scientists earn a median base near $112,590, but real offers run from about $95,000 for an entry hire to $260,000 for a principal, with total compensation at top tech employers crossing $330,000. Base pay is the easy part. What turns a $120,000 role into a $260,000 one, namely equity, specialization, and the tier of company writing the check, is where this guide spends its time.
I’m Robert Ardell, co-founder and a strategic advisor at KORE1. We have been placing technical talent since 2005, and in that stretch the data scientist title went from a rare unicorn hire to a role that sits on three different org charts at once. That fracture is the reason the salary databases struggle with the title, because they are averaging three different jobs into one number, and that average lands on top of nobody you are actually hiring. One company’s data scientist ships production fraud models next to two engineers. Another’s builds a dashboard and calls it a model. Same title. The pay gap between them runs past $80,000.
My bias, stated plainly. KORE1 places data scientists through our data scientist and data engineer staffing desk, and a fee changes hands when a client signs an offer. So a guide that talked you into a fatter band would help my side of the table. It won’t. A couple of times below I’ll point at a spot where you’re about to overpay and tell you to stop. Oversold clients do not come back. The relationships that have lasted since our early years got built on someone at KORE1 saying the uncomfortable thing early, while it still mattered.
Data Scientist Salary in 2026, at a Glance
A data scientist builds, ships, and defends models that turn messy data into a decision a business can act on. Pricing, churn, ranking, fraud, forecasting, recommendation. The title spans a career ladder five rungs deep, and pay roughly doubles from the bottom of it to the top.
The bands below composite four public sources against KORE1 placement data from the past two years, across the 30+ U.S. metros where we run data and analytics searches. Base salary only. Bonus and equity are a separate section, because for senior and staff hires the equity is often the bigger line than the raise.
| Level | Applied Experience | Base Range (US) | Total Comp at Strong Tech Employers |
|---|---|---|---|
| Entry / Junior (DS I) | 0 to 2 years | $95,000 – $120,000 | $110,000 – $145,000 |
| Mid (DS II) | 2 to 5 years | $120,000 – $155,000 | $150,000 – $210,000 |
| Senior (DS III) | 5 to 8 years | $160,000 – $210,000 | $230,000 – $330,000 |
| Staff / Principal (DS IV+) | 8+ years | $200,000 – $260,000 | $300,000 – $500,000+ |
| Data Science Manager / Lead | 6+ years | $190,000 – $250,000 | $280,000 – $450,000 |
One caveat before you screenshot that table. The total-comp column is what a funded growth-stage or public tech employer pays. A 60-person SaaS company in a second-tier metro is not paying it, and does not need to. More on that gap in a minute.
Why the Salary Trackers Disagree by Sixty Grand
Pull the same role from four sources and you get four answers. That is not the sites being sloppy. They are measuring different things, from different crowds.
The Bureau of Labor Statistics puts the 2024 median wage for data scientists at $112,590, across every employer in the country, government and insurance and retail included. It is the most neutral number you will find. It is also a base figure that does not see a dollar of stock.
Built In reports a 2026 average base of $128,067 and a median of $120,000, drawn from a more tech-heavy pool of employers. Salary.com lands close on base, with a median of $118,410. Three sources, a tight $112,000 to $128,000 cluster. So far, so calm.
Then Levels.fyi reports a median total compensation of $176,000, with the 75th percentile at $245,000 and the 90th at $330,000. Sixty thousand dollars above the others, and not wrong. Levels.fyi is fed by people who work at companies that hand out stock, so it measures total comp at the richer end of the market. BLS measures base across the whole market. Neither is lying. They are standing in different rooms.
Here is the practical version. If you are a mid-market company benchmarking a base offer, the BLS and Salary.com numbers are your reality. If you are competing with Meta or a Series C unicorn for the same candidate, the Levels.fyi number is the one walking into your interview with a counteroffer in their pocket. Budget for the room you are actually in.
Pay by Experience: First Model to Principal
The ladder is where the title earns its reputation for confusion. A junior and a principal both say “data scientist” on a badge. They are not the same hire, and the offers are not in the same area code.
Entry and junior
Fresh out of a Master’s or a bootcamp-plus-portfolio, the base sits around $95,000 to $120,000. Built In clocks the under-one-year average at $95,903, which tracks with what we see. The honest note worth saying out loud is that a lot of the titles labeled data scientist at this level are really analyst work in a fancier hoodie, and you can price them like analysts without anyone getting hurt. If the job is SQL and a dashboard, you are hiring a data analyst, and you can pay accordingly.
Mid-level
Two to five years in, owning a model end to end, the band runs $120,000 to $155,000 base. This is the most-hired rung, and it is also the one where companies overpay most often, because they reach for a senior title to win a candidate who is honestly still mid-level and a year away from earning it. The title costs you nothing today. It costs you a ladder problem in eighteen months.
Senior and above
Senior data scientists, five to eight years deep, run $160,000 to $210,000 base, and total comp at strong employers clears $300,000 once equity stacks. We cover that rung in depth in the senior data scientist salary guide. Staff and principal go higher on base and a lot higher on equity, because at that level you are paying for someone who can look at a problem and say it does not need a model, or that the data cannot answer it honestly. That call is rarer than the math, and it is what the top of the band actually buys.
Where the Job Is: Data Scientist Pay by City
Geography still moves the number, even after remote work flattened part of the map. The Bay Area pays the most and asks the most back in rent. Below are 2026 metro averages from Built In, with the national average as the baseline.
| Metro | Average Base (2026) | vs. National |
|---|---|---|
| San Francisco, CA | $172,345 | +30% |
| Remote (US) | $159,290 | +24% |
| New York City, NY | $137,473 | +12% |
| Los Angeles, CA | $133,855 | +10% |
| Seattle, WA | $133,749 | +10% |
| Boston, MA | $131,707 | +8% |
A note for the Southern California clients we work with most. Orange County roles, in Irvine, Newport Beach, and Costa Mesa, tend to land a step under the LA metro average, while still pulling Bay Area candidates who want the beach without the San Francisco rent. That arbitrage is real, and it is one of the few places a remote-friendly mid-market employer can win a senior hire on lifestyle instead of cash.
Remote pay is the line to watch. At $159,290 average, fully remote roles now out-pay every metro except the Bay. Three years ago a remote discount was standard. It mostly is not anymore, and pretending otherwise is how good offers get declined.
What Actually Moves a Data Scientist Offer
Title and city set the frame. These four things decide where inside the band you land.
Specialization. A generalist who can run a regression and a dashboard sits at the middle of the band. A data scientist who builds and ships LLM and generative AI systems, or one who does real causal inference and experimentation design, pulls $30,000 to $60,000 over that midpoint. The market is paying a steep premium for the handful of people who can make a model behave in production under real traffic, not just hit a clean accuracy score in a notebook on a quiet Tuesday afternoon.
The stack you need. Python and SQL are table stakes. Where the premium lives is the production tooling, the Spark, Databricks, Snowflake, dbt, and Airflow layer, plus the cloud platform you actually run on, whether that is AWS, Azure, or GCP. A candidate who has shipped on your exact stack saves you a six-month ramp, and they know it when they negotiate.
Industry. Finance, healthcare, and adtech pay a premium because the data is regulated, messy, or both, and a wrong model costs real money. A data scientist doing fraud at a fintech out-earns one doing marketing attribution at a mid-size retailer by a clear margin, same years of experience.
Whether they can explain it. The most underpriced skill on the market is the data scientist who can sit in front of a VP and explain why the model says what it says, and what it cannot say. That person gets promoted and counter-offered. Resumes hide it. Interviews surface it, if you ask the right way. Our guide to hiring a data scientist walks through how we screen for it.
Data Scientist, Data Engineer, ML Engineer: Who Earns What
These titles blur on job boards and the pay tracks the blur. A quick map, because hiring the wrong one against the wrong budget is the most common mistake we get called in to fix.
The data engineer builds the pipelines and warehouses that feed the models. The machine learning engineer productionizes and serves the models at scale. The data scientist frames the problem and builds the model in between. In 2026, machine learning engineers tend to edge out data scientists on base, data engineers run roughly even with data scientists, and analysts sit a clear rung below all three, which is exactly why posting the wrong title is such an expensive way to fill a seat. If you want the longer breakdown, we wrote a full data engineer vs data scientist comparison.
The practical takeaway. Do not post a data scientist req when the work is pipeline-building. You will pay a data scientist premium for engineering work, lose the candidate who wanted to model, and refill the seat in nine months.
Total Comp, Bonus, and Equity: the Part Budgets Forget
Base is the number the candidate compares first and remembers longest. It is also, above the mid-level, the smaller half of the story.
Target bonus for data scientists runs 10 to 20 percent of base at most employers, creeping higher at public tech, and it is the one piece of the package finance will usually let you flex without a board conversation. Equity is where it gets wide. At a public company, a senior data scientist’s annual stock vest can match or beat the cash bonus. At a venture-backed startup, the equity is a lottery ticket with a strike price, and a candidate who has been burned once will discount it to near zero in their head. Know which kind of equity you are offering before you quote a total-comp number, because a sophisticated candidate already has.
This is the gap that sinks offers. Finance approves a base while the candidate is comparing your total-comp story against a Levels.fyi screenshot, and if you are not putting a credible bonus and equity figure on the table in writing, you are competing on base alone against companies that are not. You can model your own bands with our salary benchmark assistant before you go to finance.
What We See Closing Offers Right Now
A few things from the desk, current to mid-2026, that the trackers lag on.
Speed wins more than money at the moment. The strong data scientists are off the market in two to three weeks, and the average time-to-hire on our IT desk runs about 17 days for a reason, namely that the clients who close fast get the candidate, and the ones running a six-week, five-round gauntlet lose them to a faster offer that was ten grand lighter. We place across direct hire and contract arrangements, and for a first data science hire a contract-to-hire start is often the lower-risk path into a permanent seat.
The other pattern. Companies keep over-leveling to win a candidate, reaching for a senior title to land someone who is genuinely still mid, and then discovering six months in that the new hire cannot actually operate at the band they are now being paid for. That is a retention problem dressed as a hiring win. KORE1’s 92% twelve-month retention rate is built on the boring opposite move, which is to level the candidate honestly, pay the band that matches, and watch the hire still be there a year later. We have run that play across 30+ U.S. metros and eight verticals since 2005.
Questions Hiring Managers and Candidates Bring Us
So what does a data scientist actually earn in 2026?
Median base is about $112,590 per the BLS, with most real offers landing between $120,000 and $210,000 depending on level. Entry hires start near $95,000; principals and big-tech total comp run past $330,000 once equity is counted.
How much should I budget for an entry-level data scientist?
Plan on $95,000 to $120,000 base for a true entry hire with 0 to 2 years. Built In pegs the under-one-year average at $95,903. If the work is mostly SQL and dashboards, you are hiring an analyst and can budget lower.
Why is the Levels.fyi number so much higher than the BLS one?
Different crowds. Levels.fyi measures total compensation at stock-granting tech companies, where its $176,000 median lives. BLS measures base pay across every employer in the country, which is why it reads $112,590. Both are accurate for the market they sample.
Do data scientists out-earn data engineers?
Roughly even on base in 2026, with machine learning engineers edging slightly ahead of both. The work decides it more than the title. Pipeline-heavy roles pay like engineering; model-and-experiment roles pay like data science.
Which city pays data scientists the most?
San Francisco, at a 2026 average base around $172,345, roughly 30 percent above the national figure. Fully remote roles now run second at about $159,290, ahead of New York, Seattle, and Boston.
What raises a data scientist’s pay the most?
Production specialization. Shipping LLM and generative AI systems or doing genuine causal inference adds $30,000 to $60,000 over a generalist. Regulated industries like finance and healthcare stack another premium on top.
Realistically, how fast can I hire one right now?
Two to three weeks for the strong candidates, who do not sit on the market. Our IT desk averages about 17 days to hire. A slow, five-round process is the single most common reason a good offer gets declined.
Is a PhD worth a higher offer?
For research-heavy and causal-inference work, yes, often $20,000 to $40,000 more. For applied product data science, a strong portfolio of shipped models matters more than the degree. Pay for the work, not the letters after the name.
How to Put This Guide to Work
Start with the room you are actually hiring in, not the most expensive screenshot you can find. Set a base band off the BLS and aggregator midpoints, then add a written bonus and equity figure if you are competing with funded tech. Level the candidate honestly. Move fast once you find the right one.
If you want a second set of eyes on a band, or a shortlist of data scientists who fit your stack and your budget, talk to a recruiter on our team. We do better when you cannot fill the seat alone, and I put that on the table in the first paragraph. We would still rather you hire the right person at the right number than the wrong person at a premium. The first one is how you become a client for fifteen years. The second is how we both lose.