Follow us

Get involved

Join or work with EvalScience

EvalScience brings together researchers and practitioners to improve how agentic systems are evaluated. There are many ways to get involved, from joining the research community to contributing expertise, resources, or support.

For individuals

Join EvalScience

We are building a network of researchers and practitioners interested in making agent evaluation more scientific, useful, and reliable.

Join a working group

Meet researchers working on agent evaluation in your area. Working groups meet regularly to discuss papers, share work in progress and start projects together.

Explore working groups and events →
Learn and train with EvalScience

Join a mentored project as a mentee or mentor, or propose a course subsection for the EvalScience open curriculum.

Go to Education →
Review, organize or contribute

Review evidence for independent assessments, organize a local community, or contribute research, briefs, or resources. These roles can range from a single event to ongoing involvement.

Join the core team

We are currently building a small core team with substantial ownership over EvalScience.

Operations & Community Lead

Run day-to-day operations, working groups, symposia, and course operations. Coordinate participants, partnerships, local organizers, funding applications, and follow-up.

Research & Network Lead

Lead the research agenda and project-matching network. Recruit mentors, track projects, maintain Research & Commons, and support research toward public outputs.

Both roles will work closely with the Founder / Executive Director and help shape EvalScience’s direction.

Tell us what you would like to contribute →

For organizations

Work with EvalScience

We work with research groups, companies, funders, and other organizations interested in improving how agentic systems are evaluated.

Research collaboration

Bring a research question your group cares about, and EvalScience can staff it through its network of collaborators, domain experts and mentored researchers. Project leads keep leadership and authorship.

Funding and partnerships

Organizations with funding or expertise can support evaluation research across the EvalScience network. Sponsors can fund compute, stipends or a full cohort of mentored projects, or support a course.

Independent evaluation and evidence review

EvalScience provides independent review and guidance for organizations seeking an external assessment of evaluation methods or claims.

For each review, EvalScience will assemble a small group of researchers and practitioners from its network with relevant expertise and no direct involvement in the work being assessed. The group will review the materials and produce a concise written assessment of what the evidence supports and what could be improved.

These reviews will generally be offered as a paid service, with reduced-cost or pro bono support available for selected academic and public-interest projects when resources allow.

Community partnership (events and working groups)

Universities, research groups, companies, and other organizations can work with EvalScience to host working groups and events, connect local researchers, or contribute to the broader evaluation community.

Support

Fund independent evaluation science

EvalScience is an independent 501(c)(3) nonprofit. Evaluation needs someone to pay for it who does not sell the systems being evaluated. Support goes to studies on our Research Agenda, mentored projects with compute and stipends for early-career researchers, an open curriculum, and events that connect researchers and practitioners across fields.

We welcome grants, gifts, compute credits, and sponsorship of a cohort, a course, or an event.

Talk to us about support

Collaboration

Current collaborators & partners

EvalScience works with researchers, research groups, and organizations across academia, industry, and the public sector. These collaborations inform our research, events, and community programs.

BRIDGE GenAI LabBRIDGE
GenAI Lab
Research collaborator

BRIDGE GenAI Lab

Beth Israel Deaconess Medical Center

Clinical AI evaluation, safety, reliability, and real-world performance.

How we work together

Co-developing evaluation methods and reusable infrastructure for clinical agent workflows.

Research collaborator

Patient Safety & Digital Health (PaDiH) Group

Danube Private University

Digital health, patient safety, medical ethics, and healthcare communication.

How we work together

Collaborating on evaluation methods and research in digital health, patient safety, medical ethics, and healthcare communication.

MTERMSResearch Lab
Research collaborator

MTERMS Lab

Brigham and Women's Hospital · Harvard Medical School

Clinical informatics, healthcare AI, natural language processing, clinical decision support and patient safety.

How we work together

Exploring scientific evaluation methods for AI systems in clinical and biomedical settings.

Raycaster
Research partner

Raycaster

Expert-generated cases and domain data for consequential AI workflows.

How we work together

Exploring how proprietary case corpora and domain expertise can support rigorous benchmark research and development.

Vector Institute
Community partner

Vector Institute

Independent research and community collaboration on AI evaluation.

How we work together

Sharing research and connecting evaluation work across the agent evaluation community.

Community partner

Digital Health and Patient Safety Platform (DHPSP)

An open research and collaboration network for digital health and patient safety.

How we work together

Connecting researchers and sharing ideas, research, and opportunities across agent evaluation, digital health, and patient safety.

Contact

Interested in collaborating or supporting EvalScience?

We welcome research collaborations, resource contributions, partnerships, and other forms of support.

Send your message directly to the EvalScience team.