Elite Engineers, Research Catalysts, Experts in Evaluation
Frontier AI research labs, AI safety research orgs, and governing bodies come to us to implement evals, verify their models, and provide information they can trust to build towards better systems.
Mission Aligned Talent
Everyone involved is here because they think measurement is a prerequisite to impact. People actually care whether projects work, not just whether they get done.
Battle Tested
Our team has deep real-world experience in frontier AI research programs. We have built industry standards in evaluation infrastructure, tooling, and environments. Our ecosystem partnerships give us a cross org context from the leaders in AI with zero fluff.
Flexible and Agile
Working with EquiStamp guarantees quality work that grows at your pace. Spin the team up or down as needed, without sacrificing outcomes. We can go from conversation to productive work in days rather than weeks.
Research Engineering, Tooling, and Verification
From eval implementation to incident investigation – we provide the full research engineering stack for clients at the frontier of intelligence. We build the tooling, environments, and measurement infrastructure that research depends on. Our services scale with industry and include experimental implementation, better information for decision makers, and continuous verification as the pace of change accelerates.
Evaluation Implementation
We build and run evaluations, baselines, and benchmarks. Working primarily with Inspect and similar frameworks, we handle the technical implementation of eval protocols.
Lifecycle Verification
Evals need to be maintained to stay relevant. We offer continuous verification, making the best information available as change impacts tools, deployments, and the world.
Red / Blue Teaming
Adversarial testing and stress-testing for AI systems. Red team finds vulnerabilities and edge cases; blue team defends and improves robustness.
Research Program Delivery
Scoping, staffing, reporting, grant administration, hiring logistics, and operational overhead. We can run the delivery layer of administrative and organizational support for your research programs.
Research Engineering
We put our engineers on your problems. Task-based work on a flexible schedule allows us to scale to fit the needs of any organization. We bring key lessons and trusted taste in research from the frontier to wider industry and vice-versa.
Incident Investigation
We help establish what happened and why when a system fails. Our independent review can provide insights on how to improve AI systems.
Core Team & Contributors
EquiStamp is run by a small core team with experience in AI safety research and operations, supported by a flexible network of technical contributors.
- Rasmus Faber-Espensen
- Rafael Carvalho
- Carlos Giudice
- Adam Hanson
- Tomáš Turlík
- Miguel Brandão
- Benjamin Shaffrey
- Shaheen Ahmed-Chowdhury
- Katalina Hernández
- Henri Lemoine
- Martin Milbradt
- Paola Baca
- Luis Slyfield
- Nikolas Mitchell
- Douw Marx
- Andrew de Wet
- Ruben Castaing
- Amy Ngo
- Samuel Prieto Lima
- Stephen Fowler
- Peter Hozák
- Myles Heller
- Matthew Goldwater
- Oscar Sykes
- Demelza Robinson
- Hannah Meagher
- Falko Galperin
- Ram Rachum
- Fraser Daly
- Qi Guo
Work With Us
For project inquiries or to join our contributor network, reach out below. We respond quickly to all serious inquiries.
Becoming a Public Benefit Corporation
EquiStamp is in the process of becoming a Public Benefit Corporation. That means our legal obligation will be to advance research, not maximize profit. When there's a choice between what's best for alignment and what's best for the bottom line, we intend to choose alignment.
Our status as a PBC enshrines our purpose while maintaining our ability to scale to meet the needs of industry and respond to market signals and incentives. It's a formal commitment to what we'd be doing anyway. We're here because we think making sure AI goes well matters, and the PBC structure will ensure that we stay true to our mission as the company grows.
Our public benefit commitment: To advance AI safety and alignment by providing essential operational and technical support that allows researchers to focus on high-impact work. We deliver evaluation implementation, data annotation, red/blue teaming, project operations, flexible research labor, and financing solutions, all while pursuing sustainable profitability. This commitment ensures our services contribute to mitigating AI risks, promoting transparent and accountable development, and enhancing global well-being through safer technologies.