For engineers and founders
What is an AI engineer?
The fastest-growing title in software: what AI engineers do, how the role differs from machine learning and software engineering, and what wage filings show it pays
An AI engineer builds products on top of AI models, usually large language models made by someone else. The work is software engineering plus the craft of making models useful and reliable, including retrieval, prompts, tool use, evaluations, and cost and latency. Machine learning engineers train and serve models, and AI engineers mostly build with them. Certified US wage filings for AI engineer titles more than tripled from fiscal 2024 to 2025.
What AI engineers do
Most AI engineers work on the layer between a model and a product. They pick which model handles which task, give it the right context through retrieval or tools, design how it’s prompted, and build the evaluations that show whether a change made things better or worse. Then they handle what any production system needs, like latency, cost, failure cases, monitoring and security.
Evaluations are what separate good AI engineers from people who have called an API. A model’s output varies from run to run, so the engineer has to build tests that measure quality across many cases and catch regressions before users do.
How it differs from related roles
- Machine learning engineer. Trains, fine-tunes and serves models, with more statistics and infrastructure in the job. AI engineers mostly use models that others have trained.
- Research engineer. Works alongside researchers on new models and training methods, usually at an AI lab.
- Software engineer. The foundation for both. Many AI engineers are software engineers who moved closer to the model as their products came to depend on it.
What it pays
We looked at certified H-1B wage filings with AI engineer titles from fiscal 2024 and 2025. Filings grew from 90 to 295 in one year. Across all 385, the median base salary was $120,000, below the $137,000 median for software engineers. Several of the most frequent filers are IT services and consulting firms, where AI engineer pay is closer to $100,000, and that pulls the national figure down. In California the median was $159,591 across 100 filings. Machine learning engineers, a longer-established title, had a median of $170,300 across 1,695 filings.
Hiring one
Look for AI features the candidate shipped that real users depend on, and ask how they knew the feature worked. Candidates who can describe their evaluation setup in detail have usually done the real work. For a startup’s first AI engineer, product judgment matters as much as model knowledge, because the hard decisions are about what the model can be trusted to do. What is a forward deployed engineer covers the customer-facing version of the role.
Common questions
- What does an AI engineer do?
- Builds products on top of AI models: choosing models, supplying context through retrieval and tools, designing prompts, building evaluations, and managing cost, latency and reliability in production.
- What is the difference between an AI engineer and a machine learning engineer?
- Machine learning engineers train, fine-tune and serve models. AI engineers mostly build products with models that others have trained.
- How much do AI engineers make?
- In certified US H-1B wage filings for fiscal 2024 and 2025, AI engineer titles had a median base salary of $120,000 nationally and $159,591 in California. Machine learning engineers had a median of $170,300.