Research Engineer vs Research Scientist

Research scientist and research engineer are useful labels, but neither has one industry-wide definition. Google DeepMind’s careers descriptions give a concrete example of the usual distinction: research scientists identify questions, develop hypotheses, design and evaluate models, and contribute to research papers; research engineers use machine-learning and software expertise to build and scale systems for testing ideas, implement and optimize models, and develop distributed-computing infrastructure. These are descriptions from one employer, not rules that govern every lab. The work can overlap substantially. Scientists may write code, build prototypes, and work on training systems; engineers may shape experiments, identify research questions, and co-author publications. OpenAI’s Research Scientist, Research Engineer, and AI Systems Engineer posting for its RSI team illustrates a shared setting where evaluations, model training, research workflows, and infrastructure meet. Team size, research area, seniority, and organization all affect who owns each task. A title alone does not establish a degree requirement, publication expectation, or level of influence. For career decisions, read the responsibilities and evidence requested in the specific posting. A role centered on new hypotheses and experimental interpretation calls for clear research examples. A role centered on scalable training, data systems, or evaluation tooling calls for strong implementation and reliability examples. Hybrid roles may expect both. Ask how the team divides experiment design, implementation, infrastructure, and publication work, and how it evaluates success. The most useful preparation follows the actual work rather than assumptions based on title.