Quick facts
- Date and time: Friday, November 20, 2026, 9:00 a.m. to 6:00 p.m. EST (America/New_York)
- Where: New York, NY. The public listing does not include a street address. More detail is pointed to evalscience.org/2026fall-nyc.
- Cost: Free
- Organizer: EvalScience. Hosts listed include Yining Hua, Stephan Rabanser, Yilun Hua, Hongbin Na, and Manuel Cherep.
- Registration: Approval required. The listing shows 162 guests. Contributions of research, tools, and failure analyses are also invited.
- Official listing: luma.com/cxmx1g4n
What it is
A one-day symposium for researchers and practitioners on what agent evaluations should measure, how to establish reliability and validity, where benchmarks and graders fail, how to build evaluation infrastructure, and what changes in interactive, real-world settings. The program is described as invited talks, contributed presentations, posters, a fireside conversation, and networking.
Featured speakers and contributors named on the page include Hanna Wallach (Microsoft Research), Jason Weston (Meta / NYU), Yoav Artzi (Cornell Tech), Nouha Dziri (Cohere Labs), Peter Henderson (Princeton), Olawale Salaudeen (UIUC), Rachneet Kaur (J.P. Morgan), Ida Momennejad (Microsoft Research), and Saurabh Jha (IBM Research). Individual talk titles are not listed on the Luma page.
Who it's for
People who already work on evaluation methods, benchmarks, agent behavior, or real-world eval infrastructure.
Worth it if / Skip it if
- Worth it if this speaker list and a full Friday on eval science are what you want, and you can get approved.
- Skip it if you need the venue address or talk titles before you commit a workday.
Verdict
Worth it for evaluation researchers and practitioners who get in. The names and the brief are strong. Confirm the site and schedule on the EvalScience page.
Source: Luma, “Is Agent Evaluation Science NYC Worth Attending? (New York, November 20, 2026)”


