Research Scientist (Neo Research)

What you'll do

• Lead research on loss-of-control and harmful manipulation in frontier models.
• Design novel evaluation methodologies, including approaches to evaluation awareness, sandbagging, and deception.
• Author safety reports and research publications.
• Set research direction in collaboration with the team.
• Engage with the wider safety community: AI safety institutes, frontier labs, academic collaborators.

What we are looking for

• Track record of original research in AI safety, evaluations, or a closely adjacent field.
• Deep familiarity with frontier model behaviour and elicitation methodology.
• Ability to define a research agenda and drive it to publication.
• Strong technical writing.
• Comfortable working closely with engineers to operationalise research.

Good to have

• PhD or equivalent independent research output.
• Published work on dangerous capability or alignment evaluations.
• Familiarity with frontier lab safety reports and regulatory frameworks (EU AI Act, NIST AI RMF).
• Reading and writing in Mandarin.

Concrete details

• Full-time. Singapore or remote.