The Title Nobody Officially Tracks
There is no official "AI engineer" occupation. The US Bureau of Labor Statistics does not collect a wage series under that name, and neither does the UK's ONS. Any single "AI engineer salary" figure you see is a vendor estimate or a self-reported average, which is why the numbers swing so wildly between sources.
The honest way to benchmark AI pay is to start from the closest occupations that are officially measured, then correct for what the market is actually posting. This guide does both: official May 2025 BLS medians, and the salary ranges disclosed in the 241 AI listings our board was carrying on 19 September 2026.
What Official US Data Shows
The closest official proxy for AI research and engineering work is "computer and information research scientists." The BLS put the median annual wage at $140,300 in May 2025. The spread is wide: the lowest 10 percent earned under $82,200, and the highest 10 percent earned over $230,630. In software publishing, the median was $211,270.
Data scientists, the other common landing spot for AI work, had a median of $120,230, with a 10th percentile of $67,240 and a 90th percentile of $199,130. Neighbouring occupations cluster around the same band:
| Closest official occupation | Median pay (May 2025) |
|---|---|
| Computer and information systems managers | $175,140 |
| Computer and information research scientists | $140,300 |
| Software developers, QA analysts and testers | $134,040 |
| Information security analysts | $129,180 |
| Data scientists | $120,230 |
| Computer systems analysts | $105,850 |
| All computer occupations | $109,470 |
| All US occupations | $50,980 |
Read together, the official picture is a median somewhere in the low-to-mid $100,000s for AI-adjacent technical work, with a long tail above $200,000 in specialised and senior roles. That is the anchor. The job market tells you where to place yourself inside it.
What 241 Live AI Listings Show
Most AI listings still hide pay. Of the 241 roles on our board on 19 September 2026, only 9 disclosed a salary range. That is about four percent, and it is the single biggest reason salary research is hard: you are comparing official medians against a market that mostly will not tell you the number.
The nine that did disclose are a useful, if small, sample of where AI pay is actually landing:
- AI Product Engineer: $129,000 to $193,500
- Data Scientist: $125,900 to $201,100
- Founding AI/Data Engineer: $120,000 to $210,000
- Senior Software Engineer, AI: $145,000 to $250,000
- AI Agent Engineer: $150,000 to $250,000
- AI Engineer, FDE Software Engineering Sr. Manager: $112,900 to $302,400
- AI Trainer, Image QA Evaluator: $30,000 to $40,000
Two things stand out. First, the technical engineering roles sit above the official computer-occupation median, often by a lot at the senior end. Second, the spread inside a single title is huge. The manager range alone runs from $112,900 to $302,400, which tells you the band is really about scope and specialisation, not the label.
Mapping the Spread to Levels
Combining the official percentiles with the disclosed listings gives a rough benchmark by level. This is our synthesis, not an official classification, and it assumes a US-based, full-time role:
| Level | Benchmark band (US) | What moves you inside it |
|---|---|---|
| Entry (0 to 2 years) | $70,000 to $110,000 | A portfolio with real evals beats coursework |
| Mid (2 to 5 years) | $110,000 to $160,000 | Owning a production system end to end |
| Senior (5 to 8 years) | $160,000 to $220,000 | Specialisation and measurable outcomes |
| Staff / lead / manager | $200,000 to $300,000+ | Scope, headcount, and scarce domain knowledge |
The most underrated lever is specialisation. The official data shows computer and information research scientists in software publishing earning a $211,270 median, far above the occupation-wide figure, for the same reason a senior AI engineer out-earns a generalist: scarcity. Skills that are hard to hire for command the premium, which is why the evaluation and agent-oversight skills we wrote about in human-in-the-loop AI jobs pay better than another generalist framework.
What Remote Really Pays
Remote does not automatically mean a pay cut, but it does mean the number is usually anchored to a location band rather than your own cost of living. A US-remote AI role often advertises a US national band, sometimes with a note that pay varies by state. Some employers tier by metro; others pay one national range and let the market sort it out.
The practical implication: when a listing says "remote" without a range, the range almost certainly exists internally. Ask for it early. It is the same transparency problem we described in why AI job listings are broken, and the fix is the same: verify before you invest time.
For remote-specific role hunting, our guide to the remote AI roles hiring now covers which titles travel well across borders, and AI job roles and functions maps the titles to the pay bands above.
How to Benchmark Your Own Number
- Start from the official proxy. Pick the BLS occupation closest to your actual work, and note its median and 10th/90th percentiles. That is your anchor.
- Adjust for specialisation. In-demand skills move you up the spread. Commodity skills move you toward the median.
- Adjust for location and remote policy. A US national band is not the same as a local-market band.
- Collect disclosed listings. Even nine ranges tell you more than a self-reported average. Track them.
- Ask early. Request the band in the first conversation. A role with no range is a role with an internal range you have not seen yet.
Conclusion
"AI engineer salary" has no single answer because the title has no official definition. The defensible method is to anchor on the closest measured occupations, where the BLS puts the relevant medians in the low-to-mid $100,000s with a long tail past $200,000, then correct for specialisation, seniority, and location using real listings.
Do that, and you can walk into a negotiation with a number you can justify, instead of a figure you found on a blog.
👉 Start with remote AI jobs or browse all current listings.