What is the difference between an applied scientist and an AI engineer?

Applied scientist is a research-flavored engineering role that originated at Amazon and is now used across tech. The job is to run experiments rigorously, choose methods that match the constraints, and produce written artifacts: papers, internal reports, model cards, evaluation memos. Applied scientists often own a single research direction for months at a time.

AI engineer is product-flavored. The job is to integrate models into systems, build the infrastructure around them, and ship. AI engineers typically own a service or a feature surface rather than a research direction.

Day-to-day, the surface overlap is high. Both read papers, both run experiments, both write code. The differences show up in success metrics. An applied scientist is judged on the credibility and impact of their published findings. An AI engineer is judged on shipped impact: latency, accuracy on a customer-facing task, revenue tied to a feature, on-call reliability.

Compensation tracks at parity for senior roles. Levels.fyi shows applied scientist L5 and L6 compensation at AWS, Apple, Meta in the same band as senior AI engineering. Frontier labs use research engineer and research scientist titles that mirror this split.

Career mobility differs in subtle ways. Applied scientist roles transition more easily into academic research or into research-focused frontier labs. AI engineering roles transition more easily into engineering management and into startup CTO tracks. Neither blocks the other; the cost of switching is one cycle of resume positioning.

Pick applied scientist if you measure your work in publishable findings. Pick AI engineer if you measure it in shipped product impact. Both are good answers.