ML Job Market 2026 — Salaries, Skills, Best Companies

The ML job market in 2026 is growing, well paid, and increasingly tilted toward experienced people. The US Bureau of Labor Statistics (BLS) projects data scientist employment to grow 35 percent between 2025 and 2035, while Indeed's Hiring Lab finds that senior roles drive most of the recent rebound in software postings. Pay data from Levels.fyi show a wide spread, from roughly $170,000 to $370,000 for the middle half of ML and AI engineers. This article puts those numbers side by side, explains why they disagree, and describes what employers actually screen for.

The 30-second picture

Three public sources describe the market from three angles. They do not agree on every number, and that is informative rather than a flaw.

Source	What it measures	Headline figure	Period
BLS Occupational Outlook Handbook	Employment and wages for a federal occupation category	Data scientists: 275,600 jobs, $120,230 median pay, 35% projected growth	Wages May 2025; growth 2025-35
Levels.fyi	Self-reported total compensation, skewed to large tech employers	ML and AI software engineers: $250,000 median; data scientists: $180,000 median	Updated October 2026
Indeed Hiring Lab	Share and count of job postings	US software development postings up almost 15% since February 2025, still about 27.5% below the pre-pandemic level	Through mid-2026

The short version: demand for people who can build and run machine learning systems is growing, pay at large technology firms is high, and the entry point has become harder. The sections below take each claim in turn.

What the official numbers say

The BLS does not publish a category called "machine learning engineer," so analysts read the market through neighboring occupations. The closest are data scientists, computer and information research scientists, and software developers.

Occupation (BLS)	Jobs, 2025	Median pay, May 2025	Projected growth, 2025-35	Openings per year
Data scientists	275,600	$120,230	35%	About 24,800
Computer and information research scientists	About 38,600	$140,300	22%	About 2,900
Software developers, QA analysts, and testers	1,905,400	$134,040	10%	About 106,100

Sources: BLS Occupational Outlook Handbook pages for data scientists, computer and information research scientists, and software developers.

A few details matter when reading this table.

Growth is not the same as openings. A 35 percent projection on a base of 275,600 jobs corresponds to about 95,400 additional positions over the decade, according to the BLS. The BLS openings figure of about 24,800 per year is larger than that decade total divided by 10 because most openings come from replacing people who change occupations or leave the workforce, not from new jobs.

The BLS wage distribution is wide. For data scientists, the lowest-paid 10 percent earned less than $67,240 and the highest-paid 10 percent earned more than $199,130. That is a three-to-one spread inside a single occupation before bonuses and equity are counted.

Education requirements differ by role. The BLS lists a bachelor's degree in mathematics, statistics, computer science, or a related field as typical for data scientists, with some employers preferring a master's or doctorate. For computer and information research scientists, a master's degree is typical and some employers prefer a Ph.D. This lines up with the experience of ML hiring: research-heavy roles lean on advanced degrees, applied roles lean on demonstrated work.

Projections are projections. The BLS builds them from historical trends and stated assumptions. They are not forecasts of any particular year, and they predate any surprise in how quickly AI tools change the work itself. Think of them as a weather forecast for a season rather than a day: useful for direction, poor for timing.

What job postings say

Official statistics lag. Job postings move faster, so researchers use them as a live indicator of demand.

AI skills are spreading beyond IT

The Stanford AI Index 2026, using Lightcast posting data and summarized by Lightcast, found that 2.5 percent of US job postings now mention AI skills. That is up 55 percent from a year earlier and nearly 300 percent from a decade ago. Lightcast also reports that most AI postings sit outside IT roles, and that the share has been rising.

Two more details from the same summary are useful for job seekers. Agentic AI skills (systems that take multi-step actions with tools) rose from 0.06 percent of postings in 2024 to 0.23 percent in 2025, nearly 90,000 US postings. And Python was the most in-demand specialized skill, appearing in 258,674 US postings, up 391 percent from the 2013 to 2015 baseline.

The implication is that "ML skills" is no longer a single job market. It is a skill that shows up in marketing analytics, finance, health care, and logistics postings, usually alongside domain knowledge.

Software postings are recovering from a low base

Indeed's Hiring Lab reported on July 8, 2026 that software development postings had risen almost 15 percent since late February 2025, while total US postings fell 7 percent over the same period. The same analysis notes that software development postings remain about 27.5 percent below their February 2020 level. So the rebound is real, and it starts from a deep hole.

The composition of the rebound matters more than its size. Seventy-one percent of the increase between May 2025 and May 2026 came from senior roles, and 37 percent came from postings with AI in the title, with the two groups overlapping. The Hiring Lab post does not cover data scientist or ML-specific postings, so these figures describe the adjacent software market rather than ML jobs directly.

The market is tilting toward seniority

A second Hiring Lab piece, published July 23, 2026, classifies postings by experience level. Senior-level positions made up 69.3 percent of software development postings in the first quarter of 2026, and entry-level positions only 4.5 percent. Across the whole US economy the mix is much more balanced, with entry-level at 46 percent, mid-level at 40 percent, and senior at about 14 percent.

Indeed defines the levels by experience: entry is roughly 0 to 1 years, mid is 2 to 4 years, and senior is 5 or more. The classification is automated, with Indeed reporting about 95 percent accuracy in manual spot checks. The finding that software-related work is far more senior-heavy than the economy as a whole is consistent with the experience of ML job seekers who report that "junior ML engineer" openings are scarce.

The entry-level squeeze, and what it does and does not prove

The most discussed labor-market finding of the last year concerns young workers. Economists Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen at the Stanford Digital Economy Lab used payroll data from ADP covering millions of US workers. In the version posted on the Stanford Digital Economy Lab's publication page, covering data through June 2026, employment of workers aged 22 to 25 in AI-exposed occupations is 19 percent below where it would have been had it kept pace with less-exposed peers. Experienced workers in the same occupations show no comparable gap.

The authors describe the results as early, descriptive indicators rather than causal estimates. Several details help interpret them:

• The effect operates mainly through reduced hiring of young workers, not through layoffs.
• It is concentrated in occupations where AI primarily substitutes for human tasks. Where AI mainly complements workers, employment is flat or rising, especially for experienced workers.
• Adjustment shows up in employment, not in base pay.
• The pattern persists when technology firms and computer occupations are excluded and when interest-rate exposure and remote work are controlled for. It weakens when education is controlled for, and some of the diverging trends predate generative AI.

For a job seeker, the sensible reading is cautious. The paper does not say that ML careers are closing. It says that employers are hiring fewer early-career people into roles where AI can substitute for entry-level tasks, and that the effect has widened since it was first documented in August 2025. Whether the cause is AI, interest rates, post-pandemic overhiring, or a mix is exactly the kind of question where correlation and causation need to be kept apart.

The practical consequence is described in the sections on skills and portfolios below: the employers who still hire early-career people are looking for evidence that you can already do part of a mid-level job.

What the pay data show

Levels.fyi collects compensation submitted by employees, mostly in technology. Its figures describe total compensation (base salary plus annualized stock plus bonus), and they skew toward large employers and toward people willing to report pay online. Use them as a map of the high end of the market rather than as the market average.

Role (US)	25th percentile	Median	75th percentile	90th percentile	Reports
ML and AI software engineer	$169,600	$250,000	$370,000	$512,000	5,945
Data scientist	$135,000	$180,000	$251,250	$340,000	3,323

Source: Levels.fyi pages for ML and AI software engineers (updated October 20, 2026) and data scientists (updated October 20, 2026). Total compensation, US.

Why Levels.fyi and the BLS disagree

The BLS median for data scientists is $120,230. The Levels.fyi median is $180,000, about 50 percent higher. Both can be correct, because they measure different things:

• The BLS surveys employers across all industries, including government, universities, and small firms. Levels.fyi is dominated by technology companies.
• The BLS measures wages. Levels.fyi includes stock and bonus, which at large public companies can be a large share of the total.
• Self-reported data over-represent people who are well paid and who track their pay.

A fair summary is that a typical data scientist in the broad economy earns something closer to the BLS figures, and that technology-sector roles at large companies can pay well above them. If you are comparing offers, ask which of those two populations you are being compared to.

Pay by location

Levels.fyi also reports medians by metropolitan area for data scientists.

Metro area	Median total compensation (data scientist)
San Francisco Bay Area	$296,000
New York City area	$200,000
Northern Virginia and Washington, DC	$152,000
Charlotte, NC area	$145,125
Greater Houston area	$143,000

Source: Levels.fyi, data scientist page, "top paying locations," updated October 20, 2026.

The Bay Area median is roughly double that of Houston. Cost of living offsets part of that gap, and the Levels.fyi data do not adjust for it. The page also lists the highest-paying employers by median total compensation: Netflix at $500,000, Stripe at $485,000, DoorDash at $450,000, Pinterest at $400,000, and Snap at $355,000. Those are medians for data scientists who reported pay at those firms, where reporting samples can be small.

The AI premium is real but shrinking at the bottom

A Levels.fyi analysis of AI engineer compensation, dated July 17, 2025, compared AI-focused software engineers with their non-AI peers at the same level.

Level	AI premium, 2024	AI premium, 2025
Entry level	10.7%	6.2%
Mid level	11.6%	11.9%
Senior	14.3%	14.2%
Staff	15.8%	18.7%

The pattern is that the premium grows with seniority. At the entry level it fell from 10.7 percent to 6.2 percent in a year, while at the staff level it rose from 15.8 percent to 18.7 percent. The post gave one striking example: at Intuit, staff-level AI engineers earned close to $917,000, compared with about $515,000 for non-AI staff engineers. That is one company, and the post does not state a sample size, so treat it as an illustration of the top of the distribution.

This matches the Indeed and Stanford findings. Both pay and hiring are concentrating among experienced people who can already work with AI systems.

What the spread means for negotiation

The distance between the 25th and 75th percentiles for ML and AI engineers, $169,600 to $370,000, is more than a factor of two. That range exists because of level, employer, location, and how much stock a package contains. It is also the space in which negotiation happens. The practical lessons are general rather than tied to any number:

• Compare offers on total compensation and on the vesting schedule of any stock, not on base salary alone. Stock is only worth something if the company's value holds.
• Ask for the level, not just the salary. Level determines the band, and a one-level difference can outweigh a negotiated bump inside a band.
• Use public data such as Levels.fyi as an input, with the caveat that it is skewed upward.
• A competing offer is generally the strongest piece of leverage, though this is a common observation rather than a measured result.

What the titles actually mean

Job titles in this field are inconsistent. The BLS describes the data scientist role as identifying useful data, collecting and cleaning it, building and validating models (including machine learning models), visualizing findings, and recommending business decisions. In practice employers use roughly these categories:

Title	Typical focus	Core tools	Hiring emphasis
Data analyst	Dashboards, reports, ad hoc analysis	SQL, BI tools such as Tableau or Power BI	Clear communication, business sense
Data scientist	Statistical analysis, experiments, predictive models	Python, SQL, statistics, machine learning libraries	Statistical judgment, stakeholder skills
ML engineer	Building, deploying, and monitoring models in production	Python plus a systems language, Docker, cloud ML services	Software engineering depth
Analytics engineer	Turning raw data into clean, tested data models	SQL, dbt (data build tool), data modeling	Data modeling and testing discipline
Research scientist	Developing new methods and publishing	Deep learning frameworks, mathematics	Advanced degree, publication record

The descriptions above are the author's summary of how the roles are commonly defined in postings, not a formal standard. The ML engineer path in particular is covered in more depth in the ML engineer career path guide, which focuses on how to get the first job, whereas this article focuses on how the market behaves.

If you are weighing these roles, a useful test is where you want to spend your hours. Analysts and data scientists spend more time on questions and evidence. ML engineers spend more time on reliability, latency, and monitoring. The two share a base in Python, SQL, and statistics, which is why people move between them.

What employers screen for

Postings and the BLS job description point to a consistent core. These are descriptive, drawn from the BLS duties list and the Lightcast skill counts, not guarantees of any employer's process.

For data scientist roles:

• Python is the most requested specialized skill in US postings, appearing in more than 258,000 of them, according to Lightcast.
• SQL appears in interviews at most companies, because data access is a daily task.
• Statistics: hypothesis testing, regression, probability, and experiment design. Mistakes here are expensive, which is why readings such as how p-values are misunderstood are worth the time.
• Machine learning fundamentals: supervised versus unsupervised learning, evaluation, and overfitting.
• Data cleaning, which the BLS lists explicitly. See what data cleaning actually involves for what that looks like in practice.

For ML engineer roles:

• Strong software engineering in Python and often a second language.
• Production practice: containers, APIs, continuous integration, and monitoring. MLOps for beginners covers the basics.
• Familiarity with at least one cloud ML platform and one experiment-tracking tool.
• For roles close to large language models, inference cost and latency awareness.

For research roles:

• Typically a master's or doctorate, consistent with the BLS statement for research scientists.
• Evidence of original work, such as papers or open-source contributions.

Portfolios

When early-career hiring is thin, evidence of work counts for more. A good portfolio has three to five projects that start from a real dataset, state a question, show the evaluation honestly, and explain a decision. One deployed project, such as a small web app or API, shows you can ship. See building a data science portfolio for examples. Certificates alone rarely substitute for this, a common hiring-manager view rather than a measured finding.

Tools

Which tools to learn is a perennial question. The best data science tools guide compares the main options, and Python versus R addresses the first choice most people face.

Where the hiring happens

Different employer types offer different trade-offs. The categories are descriptive, and the pay figures come from the Levels.fyi data above where available.

Employer type	Examples	What to expect
Frontier AI labs	OpenAI, Anthropic, Google DeepMind, Meta AI	Research-heavy, selective hiring, strong advanced-degree presence
Large technology firms	Apple, Amazon, Microsoft, Netflix	Structured levels and compensation bands; Netflix has the highest data scientist median on Levels.fyi at $500,000
Fast-growing product companies	Stripe, DoorDash, Pinterest, Snap	High reported medians for data scientists, $355,000 to $485,000
AI infrastructure	NVIDIA, Databricks, Snowflake, Hugging Face	Platform and tooling work
Enterprise software and non-tech industries	Banks, insurers, health systems, retailers	Often more stable, more domain-specific, usually lower reported pay

The Lightcast finding that most AI postings are outside IT roles suggests that the largest number of openings sit in the last row. Banks, insurers, hospitals, and manufacturers are hiring people to apply ML to fraud, risk, forecasting, and operations. Pay at these employers is more likely to resemble the BLS figures than the Levels.fyi ones, though domain expertise can command a premium.

The interview process

Processes vary by company, but a common sequence for ML and data roles is: a recruiter conversation, a technical screen covering SQL and Python, a take-home or live case, and a final round that mixes coding, ML system design, and a conversation with stakeholders. Take-homes reward clear communication over model complexity. A simple, well-justified model with an honest error analysis usually reads better than a complicated one nobody can explain.

Some preparation habits are widely reported. Know your own projects deeply enough to explain every decision. Practice SQL on realistic schemas. For ML engineer roles, rehearse a system design problem, such as a recommendation system or a fraud detector, end to end. The guide to how recommendation systems rank choices is a good example of the kind of mechanism interviewers like to probe.

Remote work

The Levels.fyi 2025 year-end report, drawing on more than 245,000 data points, found that office-based roles were the highest-growth work arrangement in its data, up 12 percent year over year. The report does not break that down for ML roles. When a remote role is offered, check whether pay is set by company headquarters, by your location, or by a national band, since that choice can move total compensation by a large margin.

Red flags in postings

Some patterns are worth questioning, based on editorial judgment rather than data:

• A posting that says "entry-level" but requires several years of experience. Indeed's definition of entry-level is 0 to 1 years, which makes such a mismatch easy to spot.
• Pay described only as "competitive" with no range, in a jurisdiction where ranges are not legally required.
• Equity offered in place of reasonable base pay at an early-stage company, without clarity on vesting and dilution.
• A very long interview process with unclear stages.
• Vague titles that bundle analyst, engineer, and scientist work into one role without clear ownership.

What could change the picture

Three uncertainties dominate the next few years.

Tools that change entry-level work. If AI coding and analysis assistants take over routine tasks, the on-ramp narrows further, as the Stanford payroll study suggests. If they instead make small teams more productive and raise the volume of ML work, demand could widen. The data so far support both stories in different places.

The business cycle. Hiring in technology tracks funding and interest rates. The Indeed data show that software postings remain well below their pre-pandemic level even while rebounding. A downturn would hit junior hiring first.

Spending on AI infrastructure. Hiring Lab's July 14, 2026 analysis of data center hiring found that data center postings have more than doubled over two years, to about 6 in every 1,000 US postings, while total US postings fell roughly 12 percent. About a quarter of those postings are for installation and maintenance work, and the 10 largest tech firms account for 71 percent of them. The post also says it is unclear how long that employment will last once construction ends. It is a reminder that AI demand creates work well outside ML job titles, and that part of it may be temporary. For a broader look at which occupations are growing, see AI jobs: the fastest-growing careers in 2026 and will AI take my job.

Career path in broad strokes

A realistic path, without attaching invented salary bands to each step:

1. Foundations (roughly the first year of focused study): Python, SQL, statistics, and the basics of machine learning. A structured plan such as the complete ML learning path helps, and the best machine learning courses compares options.
2. First role: Often an analyst, data scientist, or software engineer position. Given the entry-level data above, expect a long search and treat internships, research assistantships, and adjacent roles as legitimate ways in.
3. Mid level: Own a project from question to deployment. This is the level at which postings are most plentiful and where moving between companies is common.
4. Senior and staff: Scope grows from tasks to systems. This is where the Levels.fyi data show the AI premium widening.
5. Management or research tracks: A choice between leading people and deepening technical expertise.

Job changes are a common way to raise pay, though how much depends on timing, level, and the employer. No source cited here quantifies the typical gain, so treat any specific figure you see quoted elsewhere with caution.

Frequently asked questions

Q: Is the machine learning job market saturated? A: The data point two ways. BLS projects data scientist employment to grow 35 percent from 2025 to 2035, and Lightcast finds AI skills in 2.5 percent of US postings. But Indeed finds only 4.5 percent of software development postings are entry-level, and a Stanford study finds young workers in AI-exposed jobs are hired less. Experienced people face strong demand, while beginners face more competition.

Q: How much do ML engineers earn? A: Levels.fyi reports a US median total compensation of $250,000 for ML and AI software engineers, with a middle range of $169,600 to $370,000 (October 2026, 5,945 reports). The BLS does not track ML engineers separately; its median wage for computer and information research scientists was $140,300 in May 2025. Levels.fyi skews toward large technology employers.

Q: What does a data scientist earn? A: The BLS median was $120,230 in May 2025, with the top 10 percent above $199,130. Levels.fyi reports a median total compensation of $180,000, which includes stock and bonus and skews toward big technology firms. The two figures measure different populations, so both can be accurate.

Q: Do I need a master's degree or a PhD? A: It depends on the role. The BLS lists a bachelor's degree as typical for data scientists, with some employers preferring a master's or doctorate, and a master's as typical for computer and information research scientists. Applied roles weigh demonstrated work heavily, while research roles weigh advanced degrees and publications.

Q: Which skills matter most? A: Python is the most requested specialized skill in US postings, appearing in more than 258,000 of them, according to Lightcast. SQL, statistics, and the ability to explain results to non-specialists are consistently listed in job descriptions. Production skills such as containers, APIs, and monitoring matter for ML engineer roles.

Q: Is AI taking entry-level data jobs? A: The evidence is suggestive but not conclusive. The Stanford payroll study finds employment of 22 to 25 year olds in AI-exposed occupations 19 percent below the trend of less-exposed peers, and says the divergence works through reduced hiring. The authors call the findings descriptive, not causal, and note that some divergent trends predate generative AI.

The takeaway

The ML job market in 2026 is best described as a rising tide with a steep ramp. Official projections point to fast growth, pay at the top of the market is among the highest in any occupation, and AI skills are spreading across industries. At the same time, postings skew senior, the entry-level share in software is small, and a large payroll study finds weaker hiring of young workers in exposed jobs. For anyone planning a path, the numbers suggest treating public salary medians as a spread rather than a point, building visible evidence of work, and checking which dataset a number comes from before relying on it. This article is general information about labor-market data, not personalized career or financial advice.

Sources

• BLS Occupational Outlook Handbook: Data Scientists
• BLS Occupational Outlook Handbook: Computer and Information Research Scientists
• BLS Occupational Outlook Handbook: Software Developers, Quality Assurance Analysts, and Testers
• Levels.fyi: ML and AI Software Engineer salaries
• Levels.fyi: Data Scientist salaries
• Levels.fyi: AI Engineer Compensation Trends, Q3 2025
• Levels.fyi: 2025 End of Year Pay Report
• Indeed Hiring Lab: AI and Job Postings, From Destruction to Creation? (July 8, 2026)
• Indeed Hiring Lab: The Labor Market Is Tilting Toward Seniority (July 23, 2026)
• Indeed Hiring Lab: Hiring for the Data Center Build-Out (July 14, 2026)
• Lightcast: Four Takeaways from the 2026 Stanford AI Index
• Stanford Digital Economy Lab: Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence

Related Articles

• Build Data Science Portfolio — Projects That Get Hired
• Complete ML Learning Path 2026
• Python vs R for Data Science
• Web Development Job Market 2026 — Salaries & Careers

Last verified: October 2026