AI safety and governance

Positioning: Operations Experience → AI Safety & Governance

I come to AI safety and governance as an operator, not a researcher. As such, my value is building programs, process, infrastructure, and institutional systems. Among many challenges facing AI Safety and governance focused firms, I am most focused on three problems: the widening gap between AI capability and our ability to measure it, the concentration of advanced development inside a small number of labs, and the public's limited understanding of how these systems are actually governed. Twenty years running large diverse teams, managing hiring pipelines, building and navigating federated ecosystems, and producing public facing content at scale gives me concrete, non-metaphorical experience relevant to each.

Managing diverse, cross-disciplinary teams

Evaluating work you don't personally practice. At Yahoo Studios I managed producers, engineers, and creative talent under one P&L, and had to judge whether a technical pipeline was worth the investment, or a creative deliverable met the quality bar without being the practitioner myself. AI safety work constantly requires the same judgment: sitting between researchers and non-technical stakeholders, internalizing and translating between them, and making a defensible call without deferring blindly to either side.

Operating inside federated structures. RYOT ran with its own culture and mission inside Verizon Media's much larger infrastructure, with no monoculture imposed in either direction. Safety and governance organizations sit in a structurally similar position, whether they are semi-autonomous inside labs, discrete organizations pursuing a niche goal or funders investing in multi-stakeholder coalitions. I have a specific, lived answer to how you run something with real autonomy inside a bigger governing structure. As individual enterprises take shape, consolidate, and build steam, it is critical to architect operations to maintain the autonomy of their ideas without compromising their ability to collaborate and co-exist.

Business analysis & process development

Building measurement into fast-moving systems. The fixed/variable cost model I designed at Yahoo (lean central core, variable costs billed to business units) came from a need to accurately measure and allocate resources across output that was scaling faster than our ability to track it. That is the same operating discipline the field needs applied to model capability itself: making a fast-moving, under-measured system legible before it outruns oversight.

Building infrastructure for research-first organizations. I took RYOT, a brilliant but disorganized organization, and built the contracting, reporting, budgeting, and process structure that let it survive inside a much larger enterprise, saving $10M annually while positioning it for future sustainable growth. Safety orgs are frequently mission-driven and research-first but operationally underbuilt. My role is building the operating infrastructure (recruiting, financial, reporting, project management, communications) so the mission-critical work doesn't have to compete against operational fires.

Media & production experience

Making complex material legible to a general audience. My career has been producing editorial, branded, and live content for public consumption at scale. I know what makes a complicated topic land with a non-expert audience and what a creator or journalist actually needs to cover something accurately. That is directly relevant to closing the public understanding gap in AI governance, whether through creator engagement, field-building, or communications work.

Operating under time-boxed, irreversible pressure. Live production means real-time decisions with no undo, made alongside technical people I don't fully control. Pre-deployment evaluation and incident response work in AI safety share that shape: bounded time, real technical uncertainty, consequences that can't be walked back. This ability cannot be taught, but can only be learned through actual trial by fire experience.

What I'm focused on

Coursework

BlueDot Impact — AGI Safety Fundamentals

BlueDot Impact — AGI Strategy

BlueDot Impact — Frontier AI Governance

The governance landscape

Writing on this