Research Program Associate, AI Safety

About the role

You will partner with Research Managers and program leadership to construct and refine the operational backbone that allows fellows to conduct high-quality safety research. This position centers on designing the scaffolding for how fellows are paired with mentors, how objectives are translated into actionable plans, and how outcomes are rigorously evaluated. You will shape the procedural environment rather than directing individuals, focusing on removing obstacles and clarifying pathways. Your work will directly influence the quality of research outputs and the cohesion of the fellowship community across multiple annual cycles. You will own the definition of success metrics and iterate on them as the program scales from its initial cohort into a larger, more complex operation. You will synthesize qualitative feedback from mentors and fellows into concrete structural changes that compound in effectiveness over time.

Key facts

What you'll do

Design and iterate on core systems including fellow selection criteria, application workflows, and the definition of program deliverables that determine research quality.

Own the development and execution of an outreach campaign that identifies high-potential channels for sourcing future cohorts and maximizes effective reach.

Construct structured program evaluation frameworks that track fellow progress and program effectiveness across cohorts using measurable indicators.

Identify and implement structural improvements between fellowship cycles by synthesizing feedback from fellows, mentors, and research managers.

Support the design and execution of fellowship events such as speaker series and poster days, coordinating logistics with program leadership and external stakeholders.

Design and manage the mentor onboarding experience to ensure mentors understand expectations, have access to necessary resources, and integrate smoothly into the fellowship structure.

Develop feedback mechanisms that surface issues early and establish clear communication loops between fellows, mentors, and program leadership to prevent escalation.

Support research managers in navigating complex fellow or mentor situations that require program-level input or exception handling.

Design supplementary programming that strengthens the fellowship's intellectual environment, including reading groups, lightning talk series, and workshop sessions that complement primary research.

Establish documentation standards and processes that preserve institutional knowledge and enable scalability as the organization quadruples its team size.

Analyze data from multiple fellowship cycles to identify trends, bottlenecks, and opportunities for improving the matching process between mentors and fellows.

Coordinate logistics for multi-cycle programming, ensuring that scheduling, venue, and material requirements align with the expanding scale of the initiative.

Collaborate with leadership to translate high-level safety research goals into practical program components that can be executed consistently by non-researchers.

Continuously monitor the health of the fellowship community, intervening proactively when dynamics or structures threaten productive research environments.

Champion iterative experimentation by testing small changes in program design, measuring impact, and rolling out improvements across fellowship cohorts.

Requirements

You must have meaningful experience designing or improving programs, initiatives, or organizational systems in a research, policy, or consulting environment.

You must be able to demonstrate a track record of building processes that others find useful rather than burdensome, showing evidence of adoption and sustained use over time.

You must think in systems, naturally asking what structure would make complex work better and identifying friction points that prevent progress.

You must be an excellent communicator, capable of writing clearly, explaining concepts concisely, and delivering feedback that moves work forward without creating defensiveness.

You must be proactive when you notice problems and direct when necessary, feeling comfortable raising difficult issues with stakeholders at multiple levels.

You must be organized and conscientious, maintaining clear documentation and keeping complex, multi-stakeholder projects on schedule across overlapping cycles.

You must be receptive to feedback about your own work and capable of modifying your approach based on data, peer input, and evolving circumstances.

You must be mission-motivated by AI safety and catastrophic risk, possessing a working familiarity with AGI safety concerns and the stakes involved.

Nice to have

Experience contributing to or building communities around technical research, especially in fast-moving, high-stakes domains.

Familiarity with research mentorship models and the dynamics of advisor-advisee relationships in academic or research settings.

Prior involvement with mechanistic interpretability, model evaluation, or empirical research on advanced AI systems.

Practical notes

This is a full-time engagement with multiple fellowship cycles per year, including Fall, Spring, and Summer. Travel is not expected to be a regular component of the role. The position is based in Cambridge, Massachusetts. Candidates must be authorized to work in the United States without sponsorship for this initial cohort period.