Partnerships · Programs · Operations | AI Safety
I build the bridges across the AI safety ecosystem.
13+ years doing it at Verizon, Meta and Unreasonable Group (global network for growth-stage impact ventures): partnerships and programs, grants and events that run like clockwork.
Based in New York and Zurich, open to London or remote.
Current Lab × BlueDot Impact Fellow
Where I've worked
(global network for growth-stage impact ventures)
- UN World Food Programme
Co-founded Kintu New York (featured in Vogue Germany, Vogue Italia and Marie Claire UK).
Current work
What I’m working on now
The hardest problems in safety are often coordination problems, and coordination is my jam.
-
November 2026
Current Lab × BlueDot Impact, AI Futures Mastermind Fellowship
I built and red-teamed a theory of change for bridging AI labs and small safety organizations, so data-sharing and access agreements can get through before new models ship.
-
Now
BlueDot Impact, AGI Safety Fundamentals
-
Oct 2026 to now
AI Safety and Governance Consultant
Developed the strategy and partnership model for a university-based AI evaluation program. A field evaluation I designed for a global agricultural organization.
Previously
-
Meta
Led the global Community Ecosystem Partnerships program: 177 partnerships against a goal of 20, and directed $1.7M in Community Impact Grants, helping communities scale their impact with Meta’s tools.
-
Verizon
Led NGO and Corporate Partnerships for Verizon Small Business Digital Ready, managing 70 partners and directing grants to 60+ economic development organizations. Managed a $1.9M partnership budget and a team of 9, and moved partner management in-house, saving $300K.
-
Unreasonable Group (global network for growth-stage impact ventures)
Sold and ran $2M accelerator programs for 300 entrepreneurs, and recruited a network of 200+ mentors.
Earlier: Google; UN World Food Programme; AOL; co-founder, Kintu New York
AI safety work
A field evaluation designed for a global agricultural organization
I designed and proposed a pre-deployment field evaluation of AI advice tools, so harms can be found before anyone relies on them.
The problem
AI advice tools for low-resource users can cause harms such as wrong advice, language failures, over-trust, no recourse and privacy. Those harms need to be found before the tools are deployed.
What I designed
A pre-deployment field evaluation with participatory red-teaming, evaluation metrics, feedback and escalation loops, and partner accountability. The proposal has been submitted, and a partner meeting is scheduled.
Who it protects
Low-resource users, including farmers with low literacy and speakers of local languages.
Latest writing
From The Safety Operator
I write The Safety Operator, a newsletter about getting into AI safety from the operations, partnerships and programme side.
How I think about AI safety
The question I keep coming back to is who checks AI before it reaches the people with the least protection.
Working through a theory of change changed how I plan my career. Instead of asking where my skills are welcome, I now ask which specific risk my work reduces and where people like me are scarcest.
AI safety is a young ecosystem. Labs, independent researchers, funders, governments and civil society all hold part of the answer, but few people have the job of connecting them. Labs already employ many partnerships people. Many safety organisations, especially smaller ones, have none, so their researchers end up negotiating access, courting funders and running programmes themselves.
That’s the work I want to take on: building the bridges across the ecosystem so the right people, resources and information reach the work that needs them. Today that means agreements that get independent testers into new models before they ship, funder relationships, fellowships that bring new people in, and coalitions that let small teams be heard.
The field is moving fast, and what’s needed tomorrow may not be what’s needed today. Access is the gap now. As it changes, the bridges I build will change with it. The goal stays the same: researchers spending their time on research, and more risks caught before they reach people.
The field evaluation I designed for a global agricultural organization is one example.
Work samples
Documents and pieces I can share
Partnerships and programs
- Grant Making to Economic Development Organizations at Verizon to Scale Small Business Digitization and Innovation A partnership led model I led at Verizon to scale SMB growth on Verizon Small Business Digital Ready, including grant making to economic development organizations. Open PDF
- AI Company PE Strategy A private equity partnership strategy I developed for an AI company. Identifying details have been removed. Open PDF
- Recruiting and Growing a 200+ Mentor Network for a Global Network of 400 Growth-Stage Impact Ventures A Shell-branded recruitment briefing I created for mentors joining Shell StartUp Engine UK, with Unreasonable. Personal names and contact details have been removed. Open PDF
I am happy to serve as a reference for Sarah Nakintu, who was my direct report at Unreasonable Group. She is an exceptional connector, strategist, and GYSHIDO master (GYSHIDO: excels at getting things done). Furthermore, working with her is a truly enjoyable experience.
About
A few things to know
- Who is Sarah Nakintu?
- Sarah Nakintu builds the bridges across the AI safety ecosystem so researchers can spend their time on research. 13+ years doing it at Verizon, Meta and Unreasonable Group (global network for growth-stage impact ventures): partnerships and programs, grants and events that run like clockwork.
- What kind of AI safety work is she looking for?
- Operations, partnerships and programme roles at AI safety organizations, evaluation labs, and AI companies and labs.
- Where is she based?
- New York and Zurich, and open to London or remote.
- What is she working on now?
- She is a fellow in cohort 1 of the Current Lab × BlueDot Impact AI Futures Mastermind Fellowship in Berlin, and is taking BlueDot Impact’s AGI Safety Fundamentals course. She also works as an AI safety and governance consultant in Zurich, New York. See the field evaluation she designed for a global agricultural organization.
- What is The Safety Operator?
- The Safety Operator is her newsletter about getting into AI safety from the operations, partnerships and programme side, including a Partnerships Notes series for hiring managers.
Contact
Looking for someone to build the partnerships and programs behind your safety work? Let’s talk.
Open to operations, partnerships and programme roles at AI safety organizations, evaluation labs, and AI companies and labs.