NYU Center for Data Science

08/31/2026

AI safety testing increasingly depends on expertise that no single company can assemble on its own.

Microsoft has selected the NYU Alignment Research Group (ARG) as a member of its new External Red Team Alliance (EXTRA), a global effort to strengthen AI safety and security research.

The group received an unrestricted $25,000 gift to support its work on language models and the longer-term risks of deploying highly capable AI systems.

CDS Associate Professor of Computer Science and Data Science He He is heading up the project from the ARG side.

EXTRA connects university labs on six continents with industry practitioners.

https://www.microsoft.com/en-us/security/blog/2026/07/27/enhancing-ai-security-through-global-ai-red-teaming/

08/28/2026

Please join us in welcoming Stephen Zhang, who will join CDS as a Faculty Fellow.

Zhang is an applied mathematician who works on inverse problems, the mathematical detective work of starting from noisy, messy data and reconstructing the process that produced it, using tools from probabilistic machine learning and generative modelling.

He points those tools at questions across biology and physics, from single-cell genomics to imaging, part of the broader push toward AI for science.

We’re glad to have him.

https://nyudatascience.medium.com/meet-the-fellow-stephen-zhang-cd8549d3fc09

This entry is part of our Meet the Fellow blog series, which introduces and highlights incoming Faculty Fellows at CDS.

08/27/2026

Interested in spending the next few years doing data science in New York City?

The Fall 2027 applications for the CDS MS and PhD programs are coming soon!

CDS students work with faculty on machine learning, natural language processing, climate modeling, and the mathematics underneath all of it.

Applications for both open in late September.

In the meantime, learn more here:

MS info: https://cds.nyu.edu/masters-in-data-science-overview/

PhD info: https://cds.nyu.edu/phd-program-overview/

08/26/2026

Climate models currently run on supercomputers, run by specialists.

The NYU Earth Systems Institute is working to change that.
The cross-school hub launched in June and is led by CDS’ Laure Zanna, drawing on faculty from across NYU to predict environmental change and help people prepare for it.

The idea: build Earth system models that are faster and more accurate than today’s, run them on a single GPU, and release the weights openly so anyone can download a climate model and run it themselves.

CDS researchers are central to the effort, including Carlos Fernandez-Granda, Grace Lindsay, and incoming faculty member Zongyi Li, whose weather AI model anchors the Institute’s atmospheric modeling.

“We are training scientists to produce research, and the research and the scientists are equally important,” Zanna said.

https://nyudatascience.medium.com/climate-data-for-everyone-nyu-launches-the-earth-systems-institute-07394010df4c

Cutting-edge climate models produce useful data but currently run on supercomputers.

08/25/2026

CDS will host its Fall 2026 Graduate Career Fair on September 22, 2026, from 11:30AM-2:30PM ET. This in-person event offers employers a prime opportunity to connect with our talented data science graduate students and recent alumni.

Held in the CDS open space at 60 Fifth Avenue, the fair showcases the next generation of data scientists trained in cutting-edge techniques. Our students are seeking internships and full-time entry-level roles across various industries.

Employers can register for free, and must do so by September 11 at 5pm. For more information or to sign up, visit the link: https://docs.google.com/forms/d/e/1FAIpQLSfi0BmGbURvz_GKhy57RL4CIVNW8QuVZ-gILznfD_hoNBXnMQ/viewform

Note: This event is open exclusively to CDS students.

Questions? Contact CDS Assistant Director of Career Development Zoë Levin at [email protected].

08/21/2026

Please join us in welcoming Adam Bouyamourn, joining CDS this fall as an incoming Faculty Fellow.

A former journalist, Bouyamourn works on causal inference and on a hard question about AI: when you hand a decision to a language model, can you trust it to do what you think it is doing?

Often the people being judged start reacting to the model, and applicants rewrite their resumes to match whatever a screening tool rewards, so it ends up shaping behavior instead of measuring it.

His work runs from making automated hiring fair and accurate to understanding how governments use LLMs to monitor and suppress dissent.

We look forward to welcoming Adam to CDS.

https://nyudatascience.medium.com/meet-the-fellow-adam-bouyamourn-f97d442a9c43

This entry is part of our Meet the Fellow blog series, which introduces and highlights incoming Faculty Fellows at CDS.

08/19/2026

Language models have gotten better and better at guessing the next word in a sentence.

Oddly, that has made them worse at predicting how people actually read one.

NYU Linguistics PhD student William Timkey, working with CDS Associate Professor of Linguistics and Data Science Tal Linzen and University of Massachusetts Amherst Professor of Linguistics Brian Dillon, tracked the eyes of 368 people reading syntactically tricky sentences.

When readers moved forward, a word’s predictability explained how much they slowed down.

When readers hit a snag and looked back to reread, prediction fell short by more than tenfold.

The gap points to a second stage of reading, where we assemble a sentence’s structure and repair it after a wrong turn, that today’s models do not capture.

https://nyudatascience.medium.com/predictions-from-language-models-explain-how-we-read-a-sentence-until-we-have-to-reread-it-e389b91cf5d7

08/17/2026

The US Department of Energy has selected the first round of projects for its Genesis Mission, a national push to speed up scientific discovery with AI.

Out of more than 5,000 applications, 278 were chosen, and three involve NYU faculty, two of them from CDS.

CDS Assistant Professor of Computer Science and Data Science Mengye Ren is a co-PI on MARS (Multi-Agent Reinforcement Learning for Scientific Hypothesis Generation), led by Brookhaven National Laboratory with Carnegie Mellon University, NYU, and Microsoft.

MARS builds teams of specialized AI agents that read scientific literature, analyze experimental data, run simulations, and propose explanations, then sharpen those ideas through reinforcement learning by testing them against data.

CDS Professor of Computer Science, Engineering, and Data Science Juliana Freire is a co-PI on DAISY (Decentralized Agentic Intelligence System for Scientific Inquiry), led by Iowa State University with Argonne National Laboratory. Chinmay Hegde (VIDA, Tandon) is a co-PI.

DAISY builds a network of AI agents that divide labor and share compact knowledge tokens, with early applications planned in materials science, crop breeding for bioenergy, and the power grid.

https://www.nyu.edu/about/news-publications/news/2026/july/nyu-faculty-drive-three-us-department-of-energy-genesis-mission-.html

08/14/2026

Please join us in welcoming Xinpeng Wang, joining CDS this fall as an incoming Faculty Fellow.

Wang works on making language models more trustworthy by understanding how they work and catching failure modes like deception and reward hacking, when a model games its reward instead of doing the task it was meant to do.

That question gets harder as models take on problems where people can't easily check the answer, which is where he wants to push next.

He comes to CDS from LMU Munich, and was a visiting researcher at NYU in 2025.

We're glad to have him.

https://nyudatascience.medium.com/meet-the-fellow-xinpeng-wang-ffda015cb1c1

This entry is part of our Meet the Fellow blog series, which introduces and highlights incoming Faculty Fellows at CDS.

08/13/2026

Attention companies, nonprofits, and research labs! THIS SUNDAY, AUG 16 is the CDS Capstone Project proposal deadline!

Don’t miss your chance! Partner with CDS and tackle your data problems! Submit your proposal: https://cds.nyu.edu/masters-in-data-science-capstone/