People talk a lot about how competitive it is to hire and retain AI Engineers. But what does that actually mean? And what exactly are “AI Engineers”?
Here, we’ll explore benchmarks from Pave’s compensation data, and examine some of the nuances around AI and ML engineering.
What is the difference between AI Engineers and ML Engineers?
AI Engineers typically focus on building complete artificial intelligence systems. Machine Learning Engineers specialize in developing systems that can learn from and improve with data. AI engineering often includes ML as one of its components, while ML engineering focuses more deeply on the statistical and algorithmic aspects of training models.
That said, there's significant overlap between these roles, and many organizations use the titles interchangeably. An AI engineer at one company might have the ML engineer title at another.
AI/ML Engineering Job Architecture Benchmarks
Let’s take a look at benchmarks from Pave’s dataset, examining real-time incumbent datapoints who are explicitly classified into the ML/AL Engineer job family. Note that this is likely undercounting the total number of AI/ML Engineers, because many companies are still lumping AI/ML Engineers in the broader Software Engineering job family.
This analysis shows that, despite what the macro headlines might say, “Machine Learning” is a ~10x more common distinction than “Artificial Intelligence” for employees in the AI/ML talent pool. 83% of employees in the AI/ML job category have a title with “Machine Learning/ML”, while only 3% have a title with “Artificial Intelligence/AI”. The remaining 14% have a title with something else (e.g. “Algorithm Engineer”).
This analysis also explored some of the core categories within AI/ML roles to find the distribution of specializations. Most are not—83% map to the Generalist category. Of those that are specialized, 10% map to Applied, 4% map to Research, and 3% map to Infrastructure.
{{mid-cta}}
AI/ML Generalists vs Specialists
The vast majority of so-called “AI/ML Engineers” in the market today are likely some combination of “AI/ML Generalists" and/or “Applied AI/ML Engineers”. These folks are taking the latest frontier models and strategically applying them to their company’s tech stack to supercharge the creation of generative AI-powered features.
The Generalists/Applied AI/ML Engineers generally make a premium above the core Software Engineering job family, but it’s more in the ballpark of ~10-20% or so—rather than the order-of-magnitude sized premiums that AI/ML Researchers are clearing. A previous analysis found that AI/ML engineers also have the highest frequency of sign-on bonuses.
Given that many companies still have not broken out a standalone AI/ML job family, it is likely that the actual depth of AI/ML talent in the market today is understated, and that SWE Generalist benchmarks are being driven upwards and might be inflated to some extent.
A very small, specialized subset of the AI/ML talent pool are bona fide “Researchers”—only 4% in our dataset. These are the experts building the frontier models that are driving the GenAI wave forward. And generally speaking, these are the outliers who are bringing home comp packages measured in the millions that you read about in the news.
The Future of AI/ML Job Families
It is unclear how the job families and sub-families related to SWE, AI/ML Engineering Generalist, AI/ML Research, and so on will evolve. And this makes it particularly challenging for companies to confidently establish durable job architecture in the world of AI/ML Engineering today.
For instance, a decade from today, will all SWEs be expected to have the basic skills that the so-called AI/ML Engineering Generalists have today?
Want More Pave Data?
View cash and equity benchmarks from 8,500+ companies with Pave's free Market Data product.
Learn more about Pave’s end-to-end compensation platform
.jpeg)
