How to apply for Lead Research Scientist

About SASH

SASH is a Singapore-based nonprofit that builds AI safety research capacity and governance expertise across Asia. Neo Research (新衡) is its independent AI safety research organization studying frontier risks, with a focus on open-weight models and the growing frontier-model ecosystem in Asia. The work targets misalignment, loss of control, and harmful manipulation in increasingly capable AI systems.

About the role

The Lead Research Scientist determines what Neo should study next and builds the research programme to answer those questions. The role carries substantial ownership over individual research projects and the direction of a growing research organization. You will produce rigorous empirical evidence about risks that are important but difficult to measure.

A typical day

A typical day could involve reviewing results from an ongoing empirical study, drafting a research proposal for Neo's next project, and discussing priorities with colleagues. Since Neo is early-stage and the agenda is malleable, you may also spend time deciding which questions to pursue and what evidence standards to adopt. Exact routines are not specified, so ask about team structure and weekly cadence in interviews.

Who SASH is looking for

• Has a track record of designing and running empirical research on frontier AI risks such as misalignment, loss of control, or harmful manipulation.
• Can set a research agenda, not just execute one, and can judge which questions matter most for open-weight and Asian frontier models.
• Has experience with evaluations of model behaviour, including whether models pursue unintended objectives, conceal behaviour, recognize evaluations, evade oversight, or become less safe with greater autonomy.
• Is comfortable building research programmes and standards of evidence in an early-stage organization with a malleable agenda.

Tips for this application

• Write a research statement or proposal that names specific questions Neo should work on next and how you would measure them empirically.
• Show familiarity with open-weight models developed in Asia and explain why they are understudied from a frontier-safety perspective.
• Describe past projects where you owned the research direction, not just the execution, since this role shapes Neo's agenda.
• Address how you would set standards of evidence for risks that are hard to measure, such as concealed behaviour or evaluation recognition.
• Reference Neo Research and SASH by name and explain why a Singapore-based nonprofit focused on Asian AI safety capacity fits your work.

What to cover in your cover letter

["Your specific research agenda for Neo's next phase, with concrete questions and methods.", 'Evidence you can lead both individual projects and the direction of a growing research organization.', 'Your experience measuring hard-to-measure risks like misalignment, loss of control, or harmful manipulation.', 'Why open-weight and Asian frontier models are the right focus for your work.']

Research before applying

• Read Neo Research's published work on misalignment, loss of control, and harmful manipulation to understand current methods and findings.
• Identify which open-weight models are developed in Asia and how they are deployed globally, since this is Neo's stated focus.
• Look into SASH's work on AI safety research capacity and governance expertise across Asia to understand the parent organization.
• Check who leads Neo Research and what questions they have publicly said matter, to align your proposal with their direction.

Likely interview topics

Based on the job description, expect questions about:

• Which research questions should Neo prioritize next, and how would you decide?
• How would you design an empirical study of whether a model conceals problematic behaviour or adapts to evaluations?
• What standards of evidence should apply to claims about misalignment or loss of control?
• How do open-weight models developed in Asia differ in risk profile from closed frontier models?
• How would you build and lead a research programme in an early-stage organization with a malleable agenda?

Common mistakes to avoid

• Proposing a generic AI safety research agenda without naming specific questions or methods for Neo.
• Focusing only on closed frontier models and ignoring the open-weight and Asian ecosystem that Neo studies.
• Presenting yourself as an executor of others' research rather than someone who can set direction and standards of evidence.