Machine Learning Engineer Salary in 2026

Machine Learning Engineer Salary in 2026

Why ML Engineer Salary Numbers Vary So Widely Across Sources

If you've spent any time looking up machine learning engineer salaries, you've probably seen figures that range from $102,000¹ to $786,000² in the same afternoon. That's not noise. It's a structural problem with how salary data gets published, and it's worth understanding before you anchor to any single number.

Different surveys are measuring different things. The Bureau of Labor Statistics counts wages for broad occupational codes ("data scientists" and "computer and information research scientists") that don't map cleanly onto what a modern ML engineer actually does. Levels.fyi collects self-reported compensation from engineers at named companies, skewed toward senior US tech workers at large employers. Glassdoor aggregates anonymous self-reports across experience levels and geographies. Each approach has a real use case; none of them answers "what will I earn as an ML engineer in 2026" on its own.

Total Compensation (TC)

Total Compensation is the full annual pay figure including base salary, equity (RSUs or options, annualized over the vesting schedule), and sign-on bonus (amortized over the typical first year). TC is the number Levels.fyi leads with; base salary alone is often far lower than total comp at FAANG companies.

Base Salary

Base Salary refers to the fixed annual cash component before equity, bonus, or benefits. BLS, PayScale, and Built In report primarily on base; Levels.fyi leads with TC. Comparing a Glassdoor base figure to a Levels.fyi TC figure produces a false impression of a large pay gap that may not exist at the cash level.

The BLS May 2024 report put the median annual wage for data scientists at $112,590³, and the median for computer and information research scientists at $140,910⁴, with projected 20% employment growth from 2024 to 2034⁴. By May 2025, the BLS OEWS showed roughly 262,440 data scientists employed nationally at a median of $120,300 per year⁵. Those numbers are solid and methodologically rigorous. They're just measuring a broader population than the ML engineering roles that dominate the Levels.fyi dataset.

I'd treat the BLS figures as the floor of what the market will pay across all skill levels and geographies, and the Levels.fyi senior-tech-sector figures as the ceiling for the cohort doing production ML work at named companies. The real question is which cohort you're in.

Base Salary vs. Total Compensation: The Number That Actually Matters

The single most common mistake I see when people research ML engineer pay: comparing a base salary from one source to a total compensation figure from another. The gap isn't the market. It's the methodology.

At large tech companies, total compensation can be two to three times the base. Levels.fyi puts the median total compensation for ML engineers at $270,000, with a median of $244,500 for roles with an explicit ML/AI focus⁶. Built In, which reports primarily on base, shows an average US ML engineer base salary of $162,080⁷. Glassdoor's average lands at $162,086 per year⁸. Those figures look like they're describing different markets. They're mostly describing different components of the same offers.

Here's how to read an offer letter. The base is what hits your checking account every two weeks. RSUs vest on a schedule (typically four years with a one-year cliff), and their annual value is what Levels.fyi annualizes when it reports TC. Sign-on bonuses are one-time payments that often don't repeat in year two. When a recruiter tells you a range, ask whether that range is base or TC. At Google and Meta, the answer changes the number by a factor of two at senior levels.

RSU (Restricted Stock Unit)

An RSU is a form of equity compensation where the employer grants shares that vest over time, typically four years. The annual grant value divided by four is the common annualization used in TC calculations, though actual value fluctuates with stock price.

Sign-On Bonus

A sign-on bonus is a one-time payment made at hire, sometimes in two tranches across year 1 and year 2. Sign-ons often replace unvested equity left at a prior employer and may not repeat, so they inflate year-one TC in ways that don't persist.

For candidates at mid-sized companies or startups, equity may be options rather than RSUs, and the annualized value is harder to pin down. Built In's data showing firms with 51-200 employees paying an average of $153,309 base⁷ is more representative of what those candidates will actually see in cash. The equity upside is real but not directly comparable to a named-company RSU grant on paper.

Salary by Experience Level: Entry, Mid, and Senior

PayScale's data puts entry-level ML engineers (under 1 year of experience) at an average total compensation of $102,174, with early-career engineers (1-4 years) averaging $123,468¹. Those figures lean conservative because they include geographies and company sizes that pull the average down, but they're the most complete picture of what new entrants actually earn across the full distribution.

Motion Recruitment's 2026 compensation guide shows mid-level ML engineers in the $149,000 to $192,000 range, with senior ML engineers at $160,000 to $226,000 or more⁹. Glassdoor's data for engineers with 5 or more years of experience shows a range of $102,282 to $232,816⁸.

Two things worth noting about experience-level data:

1. "Entry-level" at a top-5 tech company is not the same as "entry-level" at a Series-A startup. The role definition, the interview bar, and the compensation structure differ sharply. PayScale's entry-level figure reflects the broad market; it doesn't tell you what Google pays a new-grad ML engineer with a strong research background.
2. The jump from mid to senior is often more about scope than years. Engineers who own production ML systems (from training pipelines through deployment monitoring) tend to exit the mid-level band faster than engineers who stay in an ML research-adjacent function without production responsibilities.

For anyone mapping their own situation: the PayScale and Motion Recruitment figures are reasonable proxies for non-FAANG roles in major metro areas. If you're targeting FAANG or frontier labs, the ceiling shifts considerably. That's what the next section covers.

Company Tier Matters More Than Title: FAANG and Frontier Lab Pay

This is the section where the numbers get genuinely large, and where I want to be careful about framing. These figures are real (they come from Levels.fyi's verified offer database), but they describe a narrow cohort. Engineers who cleared Google and Meta's interview bar in 2024-2025 are seeing a different market than engineers at mid-market companies on nominally equivalent titles.

Meta's ML engineer compensation, per Levels.fyi, runs from $187,000 total comp at the E3 (entry) level to $786,000 at E6 (staff/principal)². The median across reported Meta ML levels sits at approximately $430,000, with E6 equity grants annualizing at roughly $469,000 per year². Google's ML engineer range on the same platform goes from $199,000 at L3 to $743,000 at L7¹⁰, with a median of approximately $290,000¹⁰.

Those numbers are for engineers who cleared Google and Meta's bar. That bar is real and reproducible (thousands of engineers pass it each year), but the selection matters. If you're building toward that cohort, the path is covered in how to become an AI engineer in 2026.

KORE1's 2026 guide puts senior FAANG and frontier-lab total compensation at $350,000 and above¹¹. That's consistent with what Levels.fyi shows for E5-E6 equivalents. The spread between a senior engineer at a regional insurer and a senior engineer at Anthropic is not a modest premium. It can be $150,000 to $250,000 in annual total comp on nominally equivalent titles.

KORE1's reported base range across the full field runs $128,000 to $186,000¹¹. The base range understates the spread; the TC figures capture it. A useful heuristic: if you're evaluating an offer at a company where equity is a real portion of total pay, always ask for the full TC breakdown on the same vesting-schedule basis before comparing across offers. An offer that looks $20,000 lower in base can be $80,000 higher in annualized equity.

Glassdoor's data adds a calibration point worth keeping: large employers pay approximately 34% more than small employers for the same ML engineer title⁸. That premium is partly company-tier and partly because large companies are more likely to have established RSU programs with market-rate refresh grants.

Geography: How Much San Francisco, New York, and Seattle Add

Location still moves the number, though the remote-work normalization of 2020-2022 compressed the gap somewhat. KORE1's 2026 data shows San Francisco, New York, and Seattle paying 25 to 40 percent above the national median for ML engineers¹¹. Motion Recruitment's 2026 guide puts senior remote roles at $173,000 to $227,000⁹, which is close to, but below, senior on-site rates in the highest-cost markets.

The location-adjustment question is worth asking explicitly in any negotiation. Companies that went fully remote during 2020-2022 have been gradually implementing location-adjusted pay. Engineers hired at San Francisco rates who then relocated often found offer letters anchored to the original location; new remote hires at those same companies are increasingly seeing location-adjusted bands. That's not a universal policy, but it's a pattern that affects a real share of current candidates.

The practical read: if you're comparing an on-site offer in Seattle against a remote offer from a company headquartered in Austin, the location-adjustment policy matters as much as the headline number.

ML Engineer vs. Software Engineer: Is the Pay Gap Real?

You'll see claims that ML engineers earn thirty to forty percent more than software engineers at equivalent experience levels. The reality is more conditional.

Interview Kickstart's compensation analysis puts ML total compensation at approximately $215,000 versus approximately $175,000 for software engineers at comparable experience levels, a gap of 30 to 38 percent¹². That's consistent with a real premium for ML specialization. Still, the same analysis notes that at Google specifically, the average software engineer compensation (approximately $202,818) can slightly exceed the average ML engineer compensation (approximately $186,112) when you look at the blended level distribution¹².

That finding is worth sitting with. The ML premium is real at senior levels where ML specialization commands a scope-based step-up. It's less consistent at mid-levels at large companies with strong general SWE compensation bands. The gap is largest at applied-AI startups and frontier labs where ML is the core product, not a supporting function.

For someone deciding whether to pursue an ML-focused analytics credential versus a general software engineering track: the pay gap is real in the right context. It's not guaranteed at every company or every level.

What the Data Can't Tell You: Methodology and Its Limits

Every dataset in this article has a known bias. Being honest about that is part of how you should use these figures.

Levels.fyi is self-selected toward US tech workers at large employers who chose to share their compensation. It over-indexes on senior levels (the people most motivated to share high figures). The median TC of $270,000⁶ is not the median for the US ML engineering workforce as a whole. It's the median for the subset of ML engineers who use Levels.fyi at companies large enough to make TC worth tracking.

Glassdoor and PayScale capture more of the mid-market and small-company distribution, which is why their averages (in the $162,000 range⁷⁸) sit well below the Levels.fyi median. Neither is wrong. They're measuring different populations.

BLS data is the most methodologically rigorous because it's a census-based survey rather than self-report, but it uses occupational codes that don't isolate ML engineering cleanly. The data-scientists code (SOC 15-2051) includes roles doing descriptive analytics and business intelligence work alongside production ML. So the $120,300 median⁵ reflects a broader population than most people reading this article.

The three main methodological limits, for reference:

1. No dataset cleanly separates production ML engineering from data science, ML research, or applied-science roles that share similar titles.
2. Self-report datasets (Levels.fyi, Glassdoor, PayScale) reflect who chose to share data, not a random sample of the market.
3. Geographic scope varies: BLS is national; Levels.fyi over-represents the San Francisco Bay Area and Seattle.

What no dataset tells you: how your specific combination of skills, industry, location, and company-size preference maps to a compensation outcome. The cohort distributions above are real. Your individual result inside those distributions depends on factors these surveys don't capture (negotiation leverage, internal equity constraints at a specific company, entry path from PhD versus bootcamp, pre-IPO versus public equity). I can describe the cohort. I can't tell you where you specifically land.

Is the Switch Worth It? A Grounded Look at the ROI

The honest answer depends entirely on your starting point.

If you're a software engineer with 3 or more years of experience in Python-adjacent work, the marginal investment in ML-specific skills (systems design for training pipelines, production deployment patterns, monitoring and observability for model behavior) is relatively low given what you already have. The pay premium at companies where ML is central to the product is genuine. Whether it justifies the time depends on how far you are from the starting line. A solid first check: work through a Python for data science curriculum and see what's genuinely new versus what you already know. That gap is your actual ramp.

If you're switching from a non-technical field, the math is different. The entry-level figures from PayScale ($102,174 total comp at under 1 year)¹ are honest, but they reflect a longer ramp and often a lower starting level than experienced engineers who made a lateral move. The BLS's 20% projected job growth for computer and information research scientists through 2034⁴ suggests the demand side holds over the medium term. That projection covers a broad occupational category, not specifically the subset of the highest-paying roles.

The case for the switch is strongest when:

1. You already have adjacent technical skills (software engineering, data engineering, or statistics) that reduce the ramp time and let you compete at a higher starting level.
2. You're targeting industries or companies where ML is product-central, not a supporting function.
3. You have a specific role type in mind (production ML, research engineering, or applied science) and you've mapped the required skills against what you're actually building today.

The case is weaker when the plan is "take a bootcamp, add ML to the resume." Bootcamp-pipeline engineers are entering a more competitive market in 2026 than in 2022-2023. The pay ceiling for that cohort is real; so is the floor.

I can describe the cohort patterns above. I can't tell you whether you specifically will land where most of that cohort lands. Use this as one input alongside conversations with engineers in the actual roles you're targeting. The Levels.fyi offer data is a reasonable place to start those conversations, with the methodology caveats above in mind.

Footnotes

1. https://www.payscale.com/research/US/Job=Machine_Learning_Engineer/Salary ↩ ↩² ↩³

2. https://www.levels.fyi/companies/meta/salaries/software-engineer/title/machine-learning-engineer ↩ ↩² ↩³

3. https://www.bls.gov/ooh/math/data-scientists.htm ↩

4. https://www.bls.gov/ooh/computer-and-information-technology/computer-and-information-research-scientists.htm ↩ ↩² ↩³

5. https://www.bls.gov/news.release/ocwage.t01.htm ↩ ↩²

6. https://www.levels.fyi/t/software-engineer/title/machine-learning-engineer ↩ ↩²

7. https://builtin.com/salaries/us/machine-learning-engineer ↩ ↩² ↩³

8. https://www.glassdoor.com/Salaries/machine-learning-engineer-salary-SRCH_KO0,25.htm ↩ ↩² ↩³ ↩⁴

9. https://motionrecruitment.com/it-salary/machine-learning ↩ ↩²

10. https://www.levels.fyi/companies/google/salaries/software-engineer/title/machine-learning-engineer ↩ ↩²

11. https://www.kore1.com/ml-engineer-salary-guide/ ↩ ↩² ↩³

12. https://interviewkickstart.com/blogs/articles/machine-learning-engineer-vs-software-engineer-salary ↩ ↩²