AI Jobs: Roles, Salaries & What Recruiters Actually Want
AI work splits into two categories: jobs where AI is the subject, and jobs where it's a tool. Here's what roles exist, what they pay, and how to break in.
👁 113 reads
AI work today splits into two distinct categories: jobs where artificial intelligence is the subject of the work, and all other jobs where it has become a tool. The first group is small and well-paid, the second now covers almost every office role. Below: what roles actually exist, what they pay according to our own calculations from job listings, and what you need to do to enter that first group or at least not fall out of the second.
Salary data from job listings
We calculate this ourselves from job postings. From a sample of 202 offers with stated salary ranges, the median midpoint was $6,100 per month, with typical listings ranging from $5,500 to $6,700. The spread by seniority level is significant: mid-level around $5,000, senior around $6,100, expert around $7,200. Ranges for contractor positions are noticeably higher than for employment contracts, which you need to account for when comparing offers, as these amounts aren’t directly comparable. We update the full breakdown by area and employment type on our page how much you earn in AI; the latest calculation is from September 11, 2026.
Roles that actually exist
| Role | What you do | Where to start |
|---|---|---|
| Machine learning engineer | training and deploying models, working with data | mathematics, Python, publicly shown projects |
| AI engineer / integrator | building products on existing models, RAG, agents | programming, working with APIs, understanding costs |
| Data engineer | feeding models with data, quality and pipelines | SQL, warehouses, workflow tools |
| AI implementation specialist | translating processes into automation, training teams | n8n or Make, understanding processes, not code |
| Content and editorial with AI | creating and verifying content, quality control | writing skills and fact-checking ability |
A separate matter is the “prompt engineer” position, which in its pure form has almost disappeared from job listings. Writing prompts has become part of other roles rather than a profession, which is the natural order of things: the ability to operate a tool rarely remains a separate job for long. This is my assessment from reviewing listings, not a conclusion from market research.
What actually matters in recruitment
- Demonstrated results. One deployed project that someone uses outweighs five certificates.
- Ability to calculate costs. How much it costs to process a thousand documents, where that cost grows; few people can do this, and companies need it.
- Verification. The ability to catch when a model makes a mistake is now more important than the ability to write a prompt.
- Domain expertise. A lawyer or accountant who can use AI has a better market position than someone who only knows the tools.
Where to look for opportunities
We run our own AI job board at AI jobs; listings are free and manually approved. Beyond that: career pages of tech companies, remote job portals, and industry communities where much recruitment happens before a listing even exists.
How to prepare
If you’re starting from zero, a sensible sequence is: learn to use tools in your own work, then automate something repetitive in it, then show that result publicly. Degree programs that actually teach this field are described in the article AI degrees, and people worth knowing by name in the hub who’s who in AI.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →Common questions
Will AI take my job?
It displaces tasks, not professions: fastest are repetitive and text-based ones. Professions where responsibility for decisions and human contact matter change more slowly. This is an observation from the job listings market, not a forecast.
How much does a prompt engineer earn?
As a separate position it practically doesn’t exist. Related tasks fall into the roles listed above, with a median around $6,100 across the entire sample of offers.
Do I need a degree?
For research and training models, practically yes. For implementation, automation, and working with existing models, no; portfolio counts.
More on this from QUANTSLANT
guides·Codex vs Claude Code: Which AI Coding Agent Should You Use?OpenAI's Codex and Anthropic's Claude Code both write and test code for you, but they differ in where they run and how much control you keep.