How many AI roles are open right now and what they pay, across the US, UK and Germany. Updated weekly from live job listings.
Current openings and average advertised salary by role. Switch country to see a different market.
Salaries shown in USD, as advertised. Nothing is converted.
| Trend | |||
|---|---|---|---|
AI EngineerThe general build-with-models role: wiring models into products rather than training them. | 16,504 | $181,339 | Not enough captures |
Data ScientistIncluded as the baseline. Useful mainly as a comparison against the newer AI titles. | 5,971 | $166,459 | Not enough captures |
Machine Learning EngineerThe older, more established title. Usually implies training and deploying models, not just calling APIs. | 3,876 | $172,596 | Not enough captures |
LLM RolesRoles with LLM in the title. Narrow by design, and a good read on how specialised this work has become. | 506 | $182,896 | Not enough captures |
MLOps EngineerInfrastructure for models in production: deployment, monitoring, retraining pipelines. | 238 | $171,772 | Not enough captures |
Prompt EngineerThe title that got the most press in 2023. Worth watching precisely because it may be fading. | 96 | $125,857 | Not enough captures |
Forward Deployed EngineerEngineers who sit with the customer and build the thing on site. The best-paid title on this list. | – | – | Not enough captures |
The AI Hiring Index tracks how many AI roles are advertised and what they pay across three major markets, updated weekly from live job listings.
Job postings are a faster signal than revenue or earnings. Companies advertise roles months before the work ships, so a rising count tells you where budget has already been committed.
The gap between titles is often the most useful part. When postings for one title climb while another flattens, that is the market re-labelling the same work, and it changes who you should be recruiting.
If you are weighing a hire against outside help, the average advertised salary is only part of the cost. Recruitment, onboarding and the months before someone is productive usually add well over the headline number.
Related Tool
These are the salaries you would be competing with. If you need the work done without adding headcount, that is what we do.
What the companies doing the hiring are spending on infrastructure.
The per-token cost of the work these roles would be building.
Compare running your own hardware against cloud API spend.
Which models these teams are actually choosing between.
Yes. Each count is postings open at the time of the weekly capture, not cumulative hires or historical listings.
Many listings do not advertise a salary. When too few matching postings state one, there is no meaningful average to show and the cell is left blank rather than filled with a misleading number.
Not as raw numbers. Salaries are in local currency and nothing is converted, and total market sizes differ enormously. Compare the trend within a country rather than the level across countries.
Roughly, but not exactly. The same role can be posted to several boards and aggregated more than once, and some listings cover multiple openings. The trend is more reliable than the absolute count.
These three cover most of the readers who ask us this question. If you need another market, get in touch and we will look at adding it.
