Research Engineer vs Research Scientist SalaryRoughly equal pay at a level. The real split is the PhD bar.

Does a research scientist earn more than a research engineer at a frontier lab?

Not by much at the same level. In 2026, research engineers and research scientists at the same level at the same frontier lab are paid on broadly overlapping bands. A senior (L5-equivalent) research engineer earns total compensation of roughly $580,000 to $1,100,000, while a research scientist at the same level earns roughly $700,000 to $1,300,000 or more. The gap is a modest premium rather than a different pay class, and it is concentrated in equity for scientists with strong publication records. At staff and principal levels, both tracks regularly exceed $1,500,000 total compensation once equity is included. Some labs pay both tracks on a single uniform technical-staff band with no title-based distinction in compensation at all.

What is the difference between a research engineer and a research scientist?

A research engineer turns research ideas into runnable, scaled, reliable systems: large-scale training and evaluation infrastructure, distributed training pipelines, experiment tooling, data pipelines, and the engineering that lets a research agenda run at frontier scale. A research scientist designs the experiments, drives the technical direction, and produces the novel methodology and published findings. The two collaborate closely on the same projects, and the boundary is often fluid, but the centre of gravity differs: research engineers own the systems that make research possible, research scientists own the research questions and the results.

Do you need a PhD to be a research engineer?

No. This is the single largest practical difference between the two tracks. Research engineer roles at frontier labs are open to strong engineers without a PhD, with the bar set high but reachable through a demonstrated portfolio: major open-source projects, public model releases, reproductions of frontier results, or shipped large-scale ML systems. A meaningful share of senior research engineers do hold PhDs or equivalent published records, but it is not a gate. Research scientist roles, by contrast, are almost universally closed to non-PhD candidates, because the role is defined by the ability to conduct original research, which PhD training is the primary mechanism for developing.

Is a research engineer the same as an ML engineer?

Closely related but not identical. A research engineer works alongside research scientists on pre-production research: building the training and evaluation systems that support novel-method work, often before there is a product. An ML engineer (productisation track) focuses on shipping ML systems to users: model serving infrastructure, feature stores, experimentation frameworks, and operational reliability for systems in production. Research engineering leans upstream toward the research agenda; ML engineering leans downstream toward the product. The two markets overlap heavily, and engineers move between them, which is why research engineer compensation tracks the ML-engineer bands more closely than it tracks pure research-scientist bands.

Which track should I choose, research engineer or research scientist?

The decision rarely turns on compensation, because the pay gap at the same level is small relative to the gap between frontier-lab tier and other employer tiers. It usually rests on background and work content. If you have a PhD or a strong published research record and want to set the research agenda and be recognised as a named-author scientist, the research scientist track fits. If you are a strong engineer, with or without a PhD, who wants to build the systems that make frontier research run and prefers shipped, scaled infrastructure to publication-track recognition, the research engineer track fits and is far more accessible without an academic credential. Many engineers enter as research engineers and move toward research-scientist work over time as they build a publication record.

Can a research engineer become a research scientist?

Yes, and it is a common path, though it takes deliberate work. The transition typically requires building a published research output over one to three years: producing publishable findings, securing named authorship on research papers rather than only systems contributions, and demonstrating the ability to drive a research question rather than execute one. Research engineers who join applied research teams, where some agenda-setting is part of the role, make this transition more easily than those in pure infrastructure roles. The reverse move, research scientist to research engineer, is rarer but happens when a scientist finds systems-building more rewarding than publication-track work.

Why do some labs pay research engineers and research scientists the same?

Several frontier labs deliberately avoid a compensation distinction between the two tracks, paying both on a single uniform technical-staff band by level. The rationale is that frontier research is a tightly coupled effort in which the engineering that scales an experiment is as load-bearing as the methodology behind it, so pay is set by level and impact rather than by title. Where a premium does appear, it is usually a modest equity edge for scientists with exceptional publication records, not a structural pay-class difference. This is why the honest answer to the comparison is that the tracks pay roughly the same at a given level, and the meaningful differences are in the PhD bar, the deliverable, and who owns the research agenda.