AI Engineer Salary in 2026: What US Employers Pay, by Role, Level and City
Last updated: November 2026. Pay data: BLS May 2025 survey, released 2026.
Quick answer: the U.S. Bureau of Labor Statistics doesn't track "AI engineer" as its own job, so the honest answer comes from the occupations AI engineers are counted under. Most engineers who build AI products are counted as software developers, whose national median pay was $135,980 in May 2025, with the top 10% earning more than $214,670. Data scientists had a median of $120,230 and computer and information research scientists $140,300. Location moves the number more than anything else: the software developer median is $213,110 in San Jose and $129,940 in Las Vegas.
This guide is written for employers setting a budget for an AI hire. Every figure comes from a source listed at the end, and where a number is our own reading rather than BLS data, we say so.
Key facts
- Software developers (where most ML, LLM, AI agent and MLOps engineers are counted): median $135,980, 75th percentile $171,980, 90th percentile $214,670.
- Data scientists: median $120,230, 90th percentile $199,130. BLS projects 35% job growth from 2025 to 2035.
- Research scientists: median $140,300, 90th percentile $230,630, and 22% projected growth.
- The BLS figures leave out stock and year-end bonuses. At AI-heavy companies those can be a large part of the offer, so the true market for senior AI talent sits above these numbers.
- An employee costs more than their salary. Benefits were 30.0% of total pay for private industry workers in the second quarter of 2026, so a $171,980 salary costs about $245,686 a year.
Why There's No Official AI Engineer Salary
BLS sorts workers by the Standard Occupational Classification, and there is no "AI engineer" code in it. An employer reporting an LLM engineer to the BLS survey files them under the closest code, which is usually software developers. Research roles go under computer and information research scientists, model and analytics roles under data scientists, and leadership under computer and information systems managers.
Salary sites fill the gap with self-reported numbers or job ads. Those can be useful, but they mix countries, seniority and pay types, and they change from week to week. BLS is slower but much larger: its May 2025 estimates cover 1,687,890 software developers and 262,440 data scientists. We use it as the base, and we explain below where the real market for senior AI people runs above it.
| AI role | Closest BLS occupation | Why |
|---|---|---|
| Machine learning engineer | Software developers | Builds and ships models as production software |
| LLM and GenAI engineer | Software developers | Product engineering on top of models |
| AI agent engineer | Software developers | Backend and integration work around agents |
| MLOps and AI platform engineer | Software developers | Infrastructure and serving code |
| Data scientist, applied ML | Data scientists | Modeling, statistics and experiments |
| AI research scientist | Computer and information research scientists | New methods, often with a PhD |
| AI architect | Database architects, network architects and computer occupations, all other | BLS splits architects across several codes |
| Head of AI, VP of AI | Computer and information systems managers | Runs a technical team and budget |
The mapping is our judgment, based on how these roles are usually classified. Your own HR team may file a role differently.
AI Engineer Pay by Level
BLS publishes pay at five points across each occupation. The 10th percentile is the pay that 10% of workers earn less than. The 90th percentile is the pay that only 10% earn more than.
| BLS occupation, May 2025 | 10th | 25th | Median | 75th | 90th |
|---|---|---|---|---|---|
| Software developers | $82,460 | $105,210 | $135,980 | $171,980 | $214,670 |
| Data scientists | $67,240 | $85,660 | $120,230 | $158,880 | $199,130 |
| Computer and information research scientists | $82,200 | $103,570 | $140,300 | $188,700 | $230,630 |
| Database architects | $86,240 | $109,370 | $139,500 | $169,290 | $204,000 |
| Computer network architects | $79,900 | $104,620 | $134,050 | $168,200 | $202,680 |
| Computer occupations, all other | $55,940 | $79,370 | $116,580 | $157,500 | $188,470 |
| Computer and information systems managers | $107,550 | $138,060 | $175,140 | $220,730 | $297,510 |
How we read it for AI hiring (our judgment, not a BLS rule): a junior engineer with a year or two of real work usually lands between the 25th percentile and the median. A mid-level engineer who has shipped one production AI system sits around the median to the 75th. A senior engineer who has owned a system in production, through incidents and model changes, is at the 75th percentile or above. Staff-level and lead engineers, and anyone with rare depth like distributed training or inference optimization, start at the 90th percentile, and in the big tech metros they often pass it.
AI Engineer Pay by Role
Machine learning engineer salary
Machine learning engineers are counted as software developers, so the national median is $135,980 and the 90th percentile $214,670. Pay runs higher for engineers who own training and deployment together, because fewer people have done both. A model trained in a notebook is a data science skill. A model kept healthy in production is the one employers pay for.
LLM and generative AI engineer salary
Also counted as software developers. This is where we see the widest spread. Many candidates have built a demo on top of a model API, and far fewer have shipped a feature with an evaluation set, cost controls and a way to catch a regression when the model provider updates. Expect to pay at or above the 75th percentile ($171,980) for the second group.
AI agent engineer salary
The newest of these roles, and the one with the smallest pool of people who have shipped one to real users. It's counted as software development. Budget like a senior backend engineer with LLM depth, because most of the job is permissions, integrations and testing multi-step runs.
MLOps engineer salary
MLOps and AI platform engineers are counted as software developers too. They are paid like senior infrastructure engineers, and the premium goes to people who have run model serving under real traffic and kept GPU and inference spend inside a budget.
Data scientist salary
Median $120,230, 75th percentile $158,880, 90th percentile $199,130. Data scientists who build production models rather than reports are often retitled ML engineers and paid on the software developer scale.
AI research scientist salary
Median $140,300 and 90th percentile $230,630 across 37,200 jobs. This is a small occupation, and frontier AI labs pay well above it, mostly through equity, which BLS doesn't count.
AI architect salary
BLS spreads architects across several codes. Database architects have a median of $139,500 and a 90th percentile of $204,000. Our AI architect direct placement page goes deeper on that role, with a worked hiring cost example.
Head of AI salary
Computer and information systems managers, which includes technology leaders who run AI teams, have a median of $175,140 and a 90th percentile of $297,510. A Head of AI at a mid-market company is often hired through retained executive search, and the offer usually includes a bonus and equity that sit outside these figures.
AI Engineer Salary by City
Median pay for software developers and data scientists in 14 metro areas, May 2025. Employment is the number of people in each occupation in that metro, which tells you how deep the local pool is.
| Metro area | Software developer median | Software developers employed | Data scientist median | Data scientists employed |
|---|---|---|---|---|
| San Jose, CA | $213,110 | 87,350 | $185,080 | 6,060 |
| San Francisco and Oakland, CA | $186,640 | 69,030 | $170,110 | 10,460 |
| Seattle, WA | $167,280 | 92,770 | $164,740 | 8,370 |
| New York | $166,830 | 121,000 | $135,980 | 23,160 |
| Boston, MA | $166,090 | 42,310 | $132,040 | 7,930 |
| Los Angeles, CA | $160,920 | 55,540 | $129,740 | 9,850 |
| Washington, DC | $154,930 | 69,060 | $132,200 | 9,260 |
| Denver, CO | $137,610 | 27,010 | $112,520 | 4,510 |
| Chicago | $134,380 | 40,370 | $107,640 | 7,940 |
| Austin, TX | $134,120 | 31,960 | $127,360 | 3,730 |
| Dallas and Fort Worth, TX | $133,290 | 67,030 | $127,750 | 10,120 |
| Atlanta, GA | $132,960 | 36,300 | $108,940 | 6,820 |
| Phoenix, AZ | $131,750 | 29,380 | $114,540 | 3,480 |
| Las Vegas, NV | $129,940 | 2,900 | $98,930 | 1,180 |
Two things stand out. First, the gap between the most and least expensive metro is about $83,000 a year for the same occupation. Second, small pools change the search, not just the price. Las Vegas has 2,900 software developers, against 121,000 in New York, so a local senior AI search there usually means relocation or a remote hire. If your team is remote, decide early whether you pay one national rate or adjust by location, because candidates will ask.
What an AI Engineer Costs an Employer
Salary is only part of the cost. In the BLS employer cost survey for the second quarter of 2026, private industry employers paid $46.89 an hour in total compensation, of which $14.07 (30.0%) was benefits. Dividing a salary by 0.70 gives a rough full cost.
| Occupation | Salary level | Annual salary | Estimated cost with benefits | Cost per hour (2,080 hours) |
|---|---|---|---|---|
| Software developers | Median | $135,980 | $194,257 | $93.39 |
| 75th percentile | $171,980 | $245,686 | $118.12 | |
| 90th percentile | $214,670 | $306,671 | $147.44 | |
| Data scientists | Median | $120,230 | $171,757 | $82.58 |
| 75th percentile | $158,880 | $226,971 | $109.12 | |
| 90th percentile | $199,130 | $284,471 | $136.77 | |
| Research scientists | Median | $140,300 | $200,429 | $96.36 |
| 75th percentile | $188,700 | $269,571 | $129.60 | |
| 90th percentile | $230,630 | $329,471 | $158.40 |
The 30.0% share is an average across all private industry workers, not only technical staff, so treat these as estimates. They also leave out hiring costs: recruiting time, any agency fee, equipment, and the months the seat sits empty.
Contract AI engineers
A contract engineer is billed by the hour. On a W-2 contract the staffing firm is the employer, so the bill rate covers the engineer's pay plus employer taxes, insurance and the firm's margin, and you don't pay benefits on top. To compare a contract quote with a permanent hire, put the quote next to the "cost per hour" column above, not the salary. Our contract versus full time cost calculator runs it with your own numbers, and the staffing agency fees guide explains how markups and direct hire fees work. For AI contractors specifically, see AI staff augmentation.
What the BLS Numbers Leave Out
BLS counts base pay, commissions, incentive pay and production bonuses. Its survey FAQ lists what it excludes, including stock bonuses, year-end bonuses, profit-sharing payments, non-production bonuses, overtime and on-call pay.
For most jobs that's a small gap. For senior AI engineers it can be a big one, because large tech companies and well-funded AI startups pay a lot of the offer in equity. When you compete with them for the same person, the base salary in the table is not the number the candidate is comparing you against. You're up against the total package: base, bonus and the yearly value of the stock.
A mid-market company usually can't match a big tech equity grant, and it doesn't have to. What wins those candidates is ownership of a real system, a short and decisive interview process, and an offer that is clearly at or above the market for base pay. We tell clients early when the budget won't land the person they want, so the search doesn't burn a month first.
Is Demand for AI Engineers Still Growing?
Yes. BLS projects employment growth from 2025 to 2035 of 35% for data scientists, with about 24,800 openings a year, and 22% for computer and information research scientists, with about 2,900 openings a year. For software developers, quality assurance analysts and testers together it projects 10% growth and about 106,100 openings a year. It names AI directly as a driver, projecting strong demand "due to the continued expansion of software development for artificial intelligence (AI)."
How to Set an AI Engineer Offer That Lands
- Price the role, not the title. "AI engineer" can mean a prompt writer or someone who has run distributed training. Decide what the person must own in the first 90 days, then pick the pay band for that work.
- Use the right metro. If the role is on site, use that metro's figures. If it's remote, decide your policy before the first interview.
- Know where you sit against the 75th percentile. For a senior hire who has owned a production AI system, an offer below the 75th percentile for your metro is likely to lose to a competing offer.
- Move fast. Strong AI candidates often run several processes at once. Every extra interview round gives another company time to make an offer.
- Explain the work. Senior engineers choose the problem as much as the pay. Tell them what's in production, what's broken and what they'll own.
Our guide to how to hire AI engineers covers job descriptions and interview stages, and our AI staffing agency page covers how we run AI searches.
AI Engineer Salary FAQ
What is the average AI engineer salary in 2026?
BLS does not track AI engineers as a separate occupation. Most are counted as software developers, whose national median pay was $135,980 in May 2025, with a mean of $148,100. The top 10% earned more than $214,670.
How much do machine learning engineers make?
Machine learning engineers are counted as software developers: a median of $135,980, a 75th percentile of $171,980 and a 90th percentile of $214,670 nationally in May 2025. Engineers who own both training and production deployment are usually paid toward the top of that range.
Which US city pays AI engineers the most?
San Jose, with a software developer median of $213,110 in May 2025, followed by San Francisco and Oakland at $186,640 and Seattle at $167,280.
Do BLS salary figures include stock and bonuses?
No. BLS wage figures include base pay, commissions and production bonuses, but exclude stock bonuses, year-end bonuses, profit-sharing, overtime and on-call pay. For senior AI engineers at companies that pay in equity, the full package runs above the BLS figures.
How much does it cost to employ an AI engineer?
Benefits were 30.0% of total compensation for private industry workers in June 2026, so salary divided by 0.70 gives a rough full cost. A software developer at the $135,980 median costs about $194,257 a year, or $93.39 an hour, before hiring costs.
What does a data scientist earn?
Data scientists had a national median of $120,230 in May 2025, a 75th percentile of $158,880 and a 90th percentile of $199,130. BLS projects 35% employment growth from 2025 to 2035.
How much does a Head of AI make?
Computer and information systems managers, which includes technology leaders who run AI teams, had a median of $175,140 and a 90th percentile of $297,510 in May 2025. Leadership offers usually add a bonus and equity that BLS does not count.
Sources and Method
- National and metro pay and employment: U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics, May 2025 estimates, pulled from the BLS public data API on 3 November 2026. Occupation codes 15-1252, 15-2051, 15-1221, 15-1243, 15-1241, 15-1299 and 11-3021.
- Job growth and openings: BLS Occupational Outlook Handbook for software developers, data scientists and computer and information research scientists, projections for 2025 to 2035.
- Employer cost: BLS Employer Costs for Employee Compensation, private industry workers, June 2026.
- What wages include and exclude: BLS OEWS frequently asked questions.
- The role mapping and the reading of percentiles by level are Direcstaff's judgment from AI searches, not BLS classifications. Cost estimates divide salary by 0.70 and assume 2,080 paid hours a year.
We update this guide when BLS releases new figures. The next OEWS release, covering May 2026, is expected in 2027.
Hiring an AI engineer? Tell us the role and your budget and we'll tell you straight whether it will land the person you want, or call (702) 674-9800.