Which machine learning jobs expect a PhD?
The PhD question comes down to what the job produces. If the output is new methods, papers and experiments, you are in research scientist territory. BLS tracks this work as computer and information research scientists. Its description includes designing experiments, often with data science and machine learning techniques, and writing papers and presenting findings at conferences.
For that occupation, BLS lists a master's degree as typical entry-level education. It says some employers prefer candidates with a Ph.D., while others, such as the federal government, may hire people with a bachelor's in computer and information technology. So the PhD is a preference for some employers, according to BLS. It is not a universal rule, even in research.
Which jobs do not need one?
Machine learning engineers, AI engineers, MLOps engineers and most applied roles build, ship and maintain systems that use models. BLS has no separate AI series, so the closest match is software developers. For that occupation BLS lists a bachelor's degree as typical entry education, in computer and information technology or a related field such as engineering or mathematics. It adds that some employers prefer a master's for certain positions.
In practice, hiring for these roles turns on whether you can write production code, handle data pipelines, evaluate models and keep them running. A PhD can help with modeling depth, but it will not stand in for engineering skill.
How big is each track?
The research track is small. BLS counted about 38,600 computer and information research scientist jobs in the US in 2025, against roughly 1.9 million jobs for software developers, quality assurance analysts and testers. The federal government was the largest employer of research scientists at 31 percent, followed by computer systems design firms and R&D organizations.
BLS projects research scientist employment to grow 22 percent from 2025 to 2035, and it names artificial intelligence as one area that will need their expertise. Even with that growth, the research scientist occupation is far smaller than software development, where BLS lists a bachelor's as the typical entry education.
Should you do a PhD anyway?
Do one if you want to set research agendas, publish, or work on a research team, where some employers prefer a PhD. Skip it if you mostly want to build things. Those years can go into shipping real systems instead, and that counts for a lot on the engineering track.
A master's sits in the middle. BLS says it usually takes 2 to 3 years after a bachelor's, and it is the typical entry level for research scientists. Some developer employers also prefer it. Remember too that pay for the two tracks overlaps heavily. Engineering roles are not a lower-paid consolation prize.
Pay
USD a year, before tax. US figures. BLS rows come from OEWS May 2025, moved to 2026 with the Employment Cost Index. These are the nearest series and cover the whole occupation, not AI specialists alone. Job ad rows are midpoints of base pay ranges in 139 live AutomationRoles.ai ads from 56 employers on 5 November 2026. They lean toward large AI and tech employers and leave out stock and bonus. Refer to the Machine Learning Engineers salary guide for figures from live adverts.
What you need
- Bachelor's for most engineering rolesBLS lists a bachelor's degree as typical entry education for software developers, in computer and information technology or a related field such as engineering or mathematics.
- Master's for research rolesBLS lists a master's degree as typical entry education for computer and information research scientists.
- PhD as a preferenceBLS says some employers prefer research scientist candidates with a Ph.D. Others, including the federal government, may hire people with a bachelor's.
- Production engineering skillApplied ML work means writing and maintaining code and systems that run in production, so coding ability carries a lot of weight in hiring.
- Domain knowledge for specialist researchBLS notes that research scientists in specialized fields, such as biomedical applications, may need knowledge of that field.
How to get there
- StartEarn a bachelor's in a computing or quantitative fieldThis is the typical entry level BLS lists for software developers, the nearest series to ML engineers.
- Early careerPick a trackChoose engineering if you want to ship systems, or research if you want to create new methods and publish.
- OngoingEngineering track: build production experienceGet software or data engineering experience, then take on model training, deployment and evaluation work.
- 2 to 3 yearsResearch track: complete a master'sBLS says a master's usually takes 2 to 3 years after a bachelor's. It is the typical entry level for research scientists.
- OptionalResearch track: consider a PhDSome research employers prefer one, especially for roles that set research direction and publish.
Open now on AutomationRoles.ai
Questions people ask
Can I become a machine learning engineer without a PhD?
Yes. BLS lists a bachelor's as typical entry education for software developers, the closest series to ML engineers. Employers in this track mostly judge you on engineering and modeling skill.
Do AI research scientist jobs require a PhD?
Not always. BLS says computer and information research scientists typically need at least a master's and that some employers prefer a Ph.D. The federal government may hire some research scientists with a bachelor's.
Is a master's worth it for machine learning?
It is the typical entry level for research scientist roles, according to BLS, and some software developer employers prefer one. For applied engineering roles, experience you can show often counts for as much.
Do PhD holders earn more in machine learning?
No data here ties pay to the degree itself. The BLS median for research scientists, moved to 2026, is about $144,700, close to the $140,200 software developer median. Ad pay for ML engineers on this board runs higher, though those ads lean toward large tech employers.
Who employs most research scientists?
BLS reports that the federal government employed 31 percent of computer and information research scientists in 2025. Next came computer systems design firms at 16 percent and R&D organizations at 14 percent.
