Visit
Visit
By Grace Stanley

Yoav Artzi, associate professor of computer science at Cornell Tech and the Cornell Ann S. Bowers College of Computing and Information Science, has received the Test of Time Award as part of the inaugural Transactions of the Association for Computational Linguistics (TACL) Paper Awards. The award was announced at the ACL 2025 conference, a leading international event for research in natural language processing, which took place in Vienna, Austria, from July 27 to August 1.

The Test of Time Award recognizes research that continues to shape the field years after its publication. Artzi was honored for his 2013 paper, Weakly Supervised Learning of Semantic Parsers for Mapping Instructions to Actions, alongside co-author Luke Zettlemoyer, professor at the University of Washington.

The paper’s approach has become foundational in building AI systems that interact naturally with people, influencing technologies ranging from voice assistants to robotics. The award recognizes the paper’s lasting impact on the field of natural language understanding and its role in advancing instruction-following capabilities in artificial intelligence.

The paper introduced a method for teaching computers to understand and follow human instructions without needing detailed, manual explanations. Instead, the system learned by watching examples unfold — a form of “weak supervision” — and tracking whether its actions led to successful outcomes, like reaching a destination or completing a task.

Overall, Artzi’s paper made a significant stride in building AI that can learn from interaction and context. At Cornell Tech, Artzi’s work combines language modeling and machine learning to build AI systems that can learn from context and interact intuitively with people. His group’s research often extends into robotics, computer vision, and cognitive science.

Grace Stanley is the staff writer-editor at Cornell Tech.

< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES

By Grace Stanley

This summer, Cornell Tech welcomed the first-ever cohort of the City University of New York (CUNY) Honors Connect program, which culminated in a showcase of student-led research on July 31. The program, a collaboration between Cornell Tech and CUNY’s Macaulay Honors College, provided an opportunity for top undergraduates at CUNY colleges across New York City to work alongside Cornell Tech faculty on cutting-edge research projects.

Throughout the summer, students tackled real-world challenges across a wide range of disciplines. Their work spanned artificial intelligence, robotics, health care innovation, urban agriculture, human-computer interaction, accessibility, and intellectual property law. Whether it was designing assistive robots for emergency rooms, using virtual reality to engage young people in urban farming, or creating interactive tools to support communication for people with speech disabilities, the students helped develop innovative new technologies with direct social impact.

“The research presentations by Macaulay Honors College undergraduate students on work they conducted with Cornell Tech faculty this summer showcased why the insight of these students is critical to the vitality of our research programs,” said Greg Morrisett, Jack and Rilla Neafsey Dean and Vice Provost of Cornell Tech. “The CUNY Honors Connect program not only creates a partnership to inspire these students by exposing them to new research opportunities, but allows us to join forces to develop tech solutions for a better world.”

CUNY Honors Connect was designed to support broader undergraduate participation in STEM innovation in the heart of New York City. The program offers Macaulay Honors students a fully funded, immersive summer research experience on the Cornell Tech campus, where they are mentored by faculty and graduate researchers. Students also participated in professional development workshops, networking events, and community-building activities that prepare them for future careers in academia, industry, and entrepreneurship.

“CUNY Honors Connect exemplifies what’s possible when talented, driven students are given the opportunity to engage deeply with cutting-edge research,” said Dr. Dara N. Byrne, Dean of Macaulay Honors College. “This partnership with Cornell Tech reflects our shared commitment to expanding access to transformative, faculty-mentored experiences that prepare students to lead in the innovation economy. Together, as two of New York’s leading institutions, we’re investing in the next generation of changemakers who will shape the future of the city and beyond.”

The summer program emphasized interdisciplinary collaboration and hands-on experimentation, with close relationships with faculty and graduate researchers. The result was not only a summer of learning, but also a portfolio of projects that reflect the creativity, curiosity and dedication of the inaugural CUNY Honors Connect cohort.

Research Highlights From the 2025 Cohort

Assistive AI and Machine Learning for Health Care
Student: Areeba Ali ’28 (Macaulay at City College)

The student in assistant professor Angelique Taylor’s lab developed robotic crash carts to support emergency room staff, improving supply management and communication during medical procedures. In addition, the team explored how reinforcement learning could help hospital teams allocate tasks among healthcare workers.

In-Context Learning for Large Language Models
Student: Selina Cheng ’27 (Macaulay at Hunter College)

Under the guidance of associate professor Yoav Artzi and graduate mentor Anne Wu, the student researcher explored how generative AI can adapt to real-time feedback, pushing the boundaries of reinforcement learning and human-AI interaction.

Public Interaction With Service Robots
Students: Sehr Abrar ’26 and Elaine Huan ’28 (Macaulay at City College)

With associate professor Wendy Ju, students investigated how service robots can be better integrated into public spaces in New York City, using field experiments and large language model AI tools to analyze human-robot interactions.

Expressive Speech and CAD for Accessibility and Fabrication
Students: Krista U. Singh ’27 (Macaulay at City College) and Tri Le Minh Dinh ’26 (Macaulay at Lehman College)

In assistant professor Thijs Roumen’s lab, students worked on enhancing accessibility through AI-generated speech and humor. They also worked on a digital design tool that helps users create physical objects made from different materials and machines within a single, unified system. In an additional project with assistant professor Rajalakshmi Nandakumar, students worked on developing wireless healthcare monitoring sensors.

Urban Agriculture and Virtual Reality
Student: Samson Wu ’27 (Macaulay at City College)

Led by associate professor Tapan Parikh, the student researcher helped use virtual reality and digital storytelling to engage youth from marginalized communities in urban farming, blending technology with environmental education.

Intellectual Property and AI
Students: Hannah A. Mejia ’28 (Macaulay at City College) and Brianna Hawkins ’26 (Macaulay at Brooklyn College)

Under professor Matthew D’Amore, students examined how emerging AI technologies intersect with legal frameworks, contributing to new course materials on intellectual property law.

Wearable Ultrasound for Emboli Detection
Student: Emirosman Murtazayev ’27 (Macaulay at Baruch College)

In assistant professor Mohamed Abdelfattah’s lab, the student researcher helped prototype a wearable device capable of detecting emboli — blockages caused by blood clots or other particles that can obstruct blood flow to the brain — in real time using custom-designed electronics and machine learning.

Grace Stanley is the staff writer-editor for Cornell Tech.

< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES

By Grace Stanley

Alex Conway, assistant professor of computer science at Cornell Tech and the Cornell Ann S. Bowers College of Computing and Information Science, has won the 2025 Early Career Prize by the Society for Industrial and Applied Mathematics (SIAM) Activity Group on Applied and Computational Discrete Algorithms (ACDA). The award recognizes Conway’s influential contributions to the design and analysis of algorithms, particularly for data storage systems.

The SIAM ACDA Early Career Prize is awarded every two years to an early-career researcher who has made outstanding contributions to the field of discrete algorithms. In its announcement, SIAM highlighted Conway’s “groundbreaking contributions ranging from theory to practice, in the design, implementation, and application of algorithms and data structures for high-performance memory and storage systems.”

“My work focuses on making the fundamental building blocks of computer memory and storage systems faster and more efficient,” Conway said in the announcement. “Ultimately, this research contributes to a faster, more capable digital infrastructure, which underpins many of the technologies we rely on.”

Conway’s research at Cornell Tech focuses on bridging the gap between theoretical computer science and practical systems. One of his most notable projects is SplinterDB, a high-performance data storage system referred to as a “key-value store.” The project is now open-source and is deployed in commercial VMware products.

To celebrate his achievements, Conway will be recognized and deliver a lecture on his work at the 2025 SIAM Conference on Applied and Computational Discrete Algorithms (ACDA25), taking place from July 30 to Aug. 1 in Montreal, Canada.

“I am honored and excited to receive this award,” Conway said. “The applied and computational discrete algorithms community has been a fantastic place for exchanging ideas and building a strong network of colleagues, and I’m very much looking forward to presenting my work at the upcoming conference.”

Grace Stanley is the staff writer-editor for Cornell Tech.
< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES

By Grace Stanley

A new Cornell study revealed that Amazon’s AI shopping assistant, Rufus, gives vague or incorrect responses to users writing in some English dialects, such as African American English (AAE), especially when prompts contain typos.

The paper introduces a framework to evaluate chatbots for harms that occur when AI systems perform worse for users who speak or write in different dialects. The study has implications for the increasing number of online platforms that are incorporating chatbots based on large language models to provide services to users, the researchers said.

“Currently, chatbots may provide lower-quality responses to users who write in dialects. However, this doesn’t have to be the case,” said lead author Emma Harvey, a Ph.D. student at Cornell Tech. “If we train large language models to be robust to common dialectical features that exist outside of so-called Standard American English, we could see more equitable behavior.”

The research received a Best Paper Award at the June 23-26 ACM Conference on Fairness, Accountability, and Transparency (FAccT). Co-authors are Rene F. Kizilcec, associate professor of computer and information science at Cornell Ann S. Bowers College of Information Science, and Allison Koenecke, assistant professor at Cornell Tech.

Read more at the Cornell Chronicle. 

Grace Stanley is the staff writer-editor for Cornell Tech.

< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES

By Sarah Marquart

As tariffs shift and trade regulations grow more complex, a team of Cornell Tech students is developing a smarter solution to address modern trade compliance challenges. Their company, SAIL, is steering into uncharted waters with purpose — and a deep belief that solving the right problem can unlock a new wave of innovation.

SAIL is one of four winners at Cornell Tech’s annual Startup Awards competition on May 16. The students received a $100,000 investment, alongside mentorship and office space, from Cornell Tech.

The company’s founders are Cornell Tech graduates Chansam Kim, MBA ’25; Olivia Mei, Computer Science ’25; William J. Reid, Health Tech ’25; Salik Tehami, MBA ’25; and Ali Raza, MBA at NYU Stern.

Together, the team set out to build a solution that could streamline a critical piece of the trade compliance process: classifying goods under the Harmonized Tariff Schedule (HTS) — the global coding system that determines the duty rates applied to imported goods — and tracking duty exposure and regulatory changes in real time.

Reid’s background creating FDA-regulated machine learning tools shaped the product ethos. “We approached SAIL the same way I approached FDA-regulated machine learning: treat transparency and reliability as first-class requirements,” he said.

Cornell Tech’s Studio curriculum, a critical component of the master’s degree program, gave the team a proving ground. Over two semesters, the team leveraged its access to compliance professionals, interviewing more than 50 experts. This deep immersion helped them solidify that HTS classification as the right first problem to tackle, both in terms of urgency and scope.

“We started by listening,” said Reid. “Those interviews directly informed our decision.”

As the prototype took shape, Mei contributed her knowledge of full-stack software development. “It’s been a great opportunity to apply my skills to an important problem,” she said. “Despite a steep learning curve, collaborating with teammates who bring such diverse backgrounds and strengths has made it rewarding.”

Many of SAIL’s target users have decades of manual practice and limited exposure to AI. For that reason, SAIL positions itself as a co-pilot, not a replacement.

“We believe AI should collaborate with humans,” Kim says, capturing the startup’s ethos. With that in mind, the system explains its reasoning, cites legal rulings, and allows human reviewers to validate each output before it’s submitted.

Winning the Startup Award marked a major milestone for the team, both symbolically and strategically. “It was a great achievement,” says Kim. It gave SAIL what he calls “some certainty” — proof that the venture is much more than a class project.

The $100,000 prize will help fund in-person customer meetings, system improvements, and further development of SAIL’s AI models. It also catalyzed the team’s move into physical office space at Cornell Tech’s Roosevelt Island campus.

“It’s go time,” said Raza. “We’re soaking it in, but also recognizing that things are accelerating.”

Looking ahead, the team is focused on converting several current pilot customers into long-term contracts. Fundraising is planned for early next year, but the founders say inbound investor interest is already picking up.

“Our vision is to build the AI-native supply chain operating system for heavy industry,” says Tehami. “Most supply chain software was built for consumer goods. Heavy industry has different needs: project-based logistics, specialized equipment, and complex regulations. We’re building specifically for that reality.”

Sarah Marquart is a freelance writer for Cornell Tech.

< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES

By Henry C. Smith

The Cornell Institute for Digital Agriculture (CIDA) has announced the recipients of its 2025 Research Innovation Fund (RIF) faculty and student grants supporting new, cross-disciplinary research projects designed to improve global food systems through digital innovation. From AI-based dairy diagnostics to virtual reality for youth in agriculture, this year’s awardees are launching bold ideas that promise real-world impact.

“Solving today’s agri-food challenges demands more than innovation—it requires collaboration across disciplines, technologies, and communities,” said Fengqi You, CIDA codirector and the Roxanne E. and Michael J. Zak Professor in Energy Systems Engineering in Cornell Engineering. “The Research Innovation Fund plays a critical role in this mission by empowering faculty and students to pursue bold, early-stage ideas that can drive meaningful change from the lab to the landscape.”

Based at Cornell University, CIDA is a unique partnership that integrates fields such as engineering, computing, life sciences, veterinary medicine, and social sciences to tackle global food challenges with the help of data and digital tools.

“AI has never been more important and learning how to wield and apply AI to digital agriculture is critical,” said Hakim Weatherspoon, codirector of CIDA and professor in the Department of Computer Science in the Cornell Ann S. Bowers College of Computing and Information Science. “CIDA is a beacon of light enabling Cornell students and faculty scientists via the CIDA Research Innovation Fund to explore and advance the state of the art, improving the environment and society.”

CIDA’s RIF provides one-year seed funding for research teams that are exploring brand-new ideas in digital agriculture. The awards help faculty gather early data, build new collaborations, and prepare for larger national or global grants.

Grants are divided into two categories:

  • Planning Grants (up to $15,000): For developing big, early ideas into large-scale proposals.
  • Seed Grants (up to $30,000): For testing innovative solutions with potential for high impact.

 

CIDA integrates fundamental discoveries across the agricultural and life sciences, engineering, computing and information sciences, and social sciences to advance equitable, sustainable, and efficient agriculture and food systems.

To qualify, each project must involve at least two faculty members from different Cornell colleges and represent new collaborations not previously funded by CIDA. Awardees will present at the October 21, 2025 CIDA workshop (registration links will be found at the CIDA News + Events page), as well as giving a presentation in the newly launched Cornell course AI for Digital Agriculture.

“Supporting new cross-disciplinary collaborations among researchers in diverse disciplines from different colleges is at the core of CIDA’s mission,” said Julio Giordano, codirector of CIDA and professor of animal science in the College of Agriculture and Life Sciences (CALS). “These RIF-funded projects exemplify the potential transformative value of innovation across engineering, data science, and agriculture.”

2025 CIDA RIF Awarded Projects

Faculty Planning/Seed Grant Projects

Frameworks for Ethical Agricultural Data Aggregation – Sara Emery, Renata Ivanek (College of Veterinary Medicine), Steven Wolf (CALS), Katie Gold (CALS)

Biosensor Tattoos for Livestock Health – Taika von Königslöw (CVM), Cindy Hsin-Liu Kao (College of Human Ecology), Jennifer Sun (Ann S. Bowers College of Computing and Information Science)

Using VR to Increase Youth & Minority Participation in Agriculture – Tapan Parikh (Cornell Tech/CIS), Jenny Kao-Kniffin (CALS)

Maintaining Agricultural Machinery – Thijs Roumen (Cornell Tech/CIS), Francois Guimbretiere (CIS), Terry Bates (CALS), Elizabeth Bihn (CALS)

Visualizing Water Movement in Root Zones – Taryn Bauerle (CALS), Amit Lal (College of Engineering)

Reducing Methane and Arsenic in NY Rice Paddies – Matthew Reid (COE), Jenny Kao-Kniffin (CALS), Chuan Liao (CALS)

CIDA’s RIF also funds Student Summer Research Awards (up to $11,310) for innovative research projects in digital agriculture under the mentorship of two faculty members from different Cornell colleges.

“The beauty of the CIDA Student RIF Program lies in the rich diversity of student backgrounds and the wide range of challenges being addressed—where innovations in AI, data science, agriculture, biology, and health converge to advance sustainable food production, from dairy calf diagnostics to grapevine genomics,” said Renata Ivanek, codirector of CIDA and professor of population medicine and diagnostic sciences in the College of Veterinary Medicine.

Student Awards

Automated Pose Estimation in Surplus Dairy Calves – Ryan Ye (CALS Undergraduate), Jennifer Sun (CIS), Francisco Leal-Yepes (CVM)

Smart Ultrasound for Diagnosing Calf Respiratory Illnesses – Marina Madureira Ferreira (CALS Graduate Student), Francisco Leal-Yepes (CVM), Jennifer Sun (CIS)

Modeling H5N1 Transmission in Dairy Cattle – Zijun He (CVM Graduate Student), Renata Ivanek (CVM), Jennifer Sun (CIS)

AI Strategies for Sustainable Greenhouse Farming – Jinsung Kim (COE Graduate Student), Fengqi You (COE), Yu Jiang (CALS)

Standardizing Dairy Farm Treatment Records with LLMs – Xinyu Yang (CIS Graduate Student), Jennifer Sun (CIS), Renata Ivanek (CVM)

SplatOverflow: Machine Training Aid – Amritansh Kwatra (Cornell Tech/CIS Graduate Student), Thijs Roumen (Cornell Tech/CIS), Elizabeth Bihn (CALS)

LLM Agent for Grapevine Disease Resistance Research – Daoyuan Jin (COE Graduate Student), Yu Jiang (CALS), Jennifer Sun (CIS)

Validating GrapeSAM Vision Model for Wine Grapes – Angela Paul (CALS Graduate Student), Justine Vanden Hueuvel (CALS), Jonathan Jaramillo (COE)

Sorghum Trait Prediction Using NIR Spectroscopy – Marjorie Hanneman (CALS Graduate Student), Mike Gore (CALS), Abe Davis (CIS)

Henry C. Smith is the communications specialist for Biological Systems at Cornell Research and Innovation.

 

< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES

By Grace Stanley

Cornell Tech has announced the appointment of Fei Wang, professor at Weill Cornell Medicine, as the new Senior Faculty Fellow in Clinical AI. The newly established role will strengthen Cornell Tech’s leadership in digital health and artificial intelligence, while also expanding interdisciplinary collaboration between Cornell Tech and Weill Cornell Medicine.

In his new position, Wang will lead efforts to expand research and development in clinical AI across the two campuses. His responsibilities also include working with Cornell Tech’s Health Tech Hub on clinical AI strategy, mentoring faculty and researchers on AI-driven health initiatives, and teaching and advising master’s students at Cornell Tech.

“I am beyond excited to become an official member of the Cornell Tech family,” Wang said. “I’m committed to pushing the boundaries of research and education at Cornell Tech while building even stronger ties with Weill Cornell Medicine.”

Wang brings a distinguished track record in health informatics and AI. He was recently appointed Chief of the Division of Health Informatics and Artificial Intelligence and Associate Dean for Artificial Intelligence and Data Science at Weill Cornell Medicine. He also serves as senior technical advisor at NewYork–Presbyterian Hospital.

Since joining Weill Cornell Medicine in 2016, Wang has played a pivotal role in advancing AI applications in health care. He is a Cornell University field member in computer science, information science, electrical and computer engineering, statistics and data science, and computational biology. His research has focused on developing machine learning models for biomedical data analysis, such as patient similarity analytics, disease subtyping, and clinical trial emulation.

Wang is also the founder of the Institute of AI for Digital Health at Weill Cornell Medicine, which brings together experts in computational science and clinical medicine to develop AI technologies that support individualized health care. The institute collaborates with Cornell Tech, NewYork-Presbyterian Hospital, the Hospital for Special Surgery, and Rockefeller University.

“Fei has already made a tremendous impact through his collaborations with Cornell Tech faculty and students,” said Tanzeem Choudhury, Roger and Joelle Burnell Professor in Integrated Health and Technology at Cornell Tech and and the Jacobs Technion-Cornell Institute. “His deep expertise in clinical AI and his passion for mentorship have enriched our community. We’re excited to see how his leadership will help us scale our efforts and drive meaningful innovation in health care.”

A graduate of Tsinghua University, where he earned his doctoral degree in machine learning, Wang is an elected fellow of the American College of Medical Informatics, the International Academy of Health Sciences Informatics, and the American Medical Informatics Association. He is also a distinguished member of the Association for Computing Machinery.

Wang’s work has earned numerous accolades, including the NSF CAREER Award, the IEEE International Conference on Health Informatics Research Leadership Award, the Google Faculty Research Award, the Amazon Machine Learning for Research Award, and the Sanofi iDEA Award. He holds more than 30 patents related to AI and machine learning in health care.

Grace Stanley is the staff writer-editor for Cornell Tech.

< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES

By Grace Stanley

Cornell Tech assistant professor Raaz Dwivedi has co-founded Traversal, a startup emerging from stealth this week with a mission: to transform how modern software systems detect and resolve outages using artificial intelligence.

The company has raised a $48 million seed and Series A round led by Sequoia Capital and Kleiner Perkins, with participation from NFDG and Hanabi. With headquarters in New York City, Traversal embodies the Cornell Tech ethos: bridging research, innovation, and real-world impact, right in the heart of the city.

“The company’s growth was fueled by all of us being in New York at the same time. The velocity you hit when you’re in person — I think it accelerated our development. And New York’s burgeoning startup scene is energizing,” said Dwivedi, who is also affiliated with Cornell Engineering.

Dwivedi credits Cornell Tech’s entrepreneurial ecosystem as a key enabler. “Cornell Tech’s support for external engagement made it the perfect place to take that leap,” he said. “From day one, I felt that culture of innovation and entrepreneurship — and it gave me the confidence to start something ambitious.”

Traversal is building what it calls an “AI SRE,” a 24/7 intelligent companion for site reliability engineering. The platform continuously monitors software systems to identify and troubleshoot complex production incidents. By leveraging telemetry — including logs, metrics, traces, and code changes — it pinpoints what Dwivedi calls the “smoking gun” behind outages. But Traversal goes a step further, identifying potential issues before they escalate.

“We’re quickly moving toward self-healing systems,” Dwivedi explained. “AI can detect what’s broken, identify the root cause, and suggest, or even implement, a fix within minutes.”

In practice, this can significantly reduce downtime and ease engineering stress, particularly at large enterprises with complex infrastructure and fragmented data. Traversal is already being used by Fortune 500 companies, including mission-critical cloud providers and financial institutions.

Dwivedi’s expertise in causal machine learning and reinforcement learning has been foundational to solving this problem.

“Root cause analysis is fundamentally a search problem,” he said. “At the heart of it, it’s about finding a needle in a very large haystack (or dozens of haystacks) with many fake needles everywhere. Given the recent advancements in large language models and causal machine learning, I think this is the perfect time to build a semantics-meets-statistics framework to solve this problem. And Traversal is doing just that.”

He co-founded Traversal with fellow AI researchers Anish Agarwal, Raj Agarwal, and Ahmed Lone. The team’s roots trace back to their time at UC Berkeley, MIT, and Columbia.

“We were all researchers, but this felt like the right moment to translate our work in AI into something impactful for industry,” said Dwivedi.

Grace Stanley is the staff writer-editor for Cornell Tech.

< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES

By Sarah Marquart

Medical crash carts — stocked with medications, syringes, gauze, and IV fluids — are essential tools in emergency rooms, where seconds matter, space is tight, and confusion can cost lives. However, despite their importance, crash carts vary widely in layout between hospitals and departments. In high-stress situations, this inconsistency can lead to delays as providers search for supplies or open the wrong drawers.

That’s why a team of Cornell researchers set out to reimagine the traditional crash cart, transforming it into a robot designed to support, not disrupt, emergency care teams during critical, time-sensitive procedures.

Their work, led by Angelique Taylor, assistant professor of information science at Cornell Tech, was presented in February 2025 at the ACM/IEEE International Conference on Human-Robot Interaction (HRI). The team’s paper, “Rapidly Built Medical Crash Cart! Lessons Learned and Impacts on High-Stakes Team Collaboration in the Emergency Room,” shares insights from a multi-year effort to design, prototype, and evaluate a robotic, remotely-operated version of the traditional crash cart.

As Taylor explains, the goal wasn’t to introduce entirely new equipment, but to rethink how existing tools could better support teamwork in high-pressure care settings.

“The medical crash cart robot is a cart-based platform designed to support team collaboration during medical procedures, specifically resuscitation code procedures in Emergency Departments,” explained Taylor, who is also affiliated with the Cornell Ann S. Bowers College of Computing and Information Sciences (Bowers CIS). “We wanted to explore how existing equipment in the room can be used to support team collaboration.”

To understand how this robot could function in real-life scenarios, the team designed and tested three versions of the cart-based platform. Each prototype was designed to address a core question: How can a robot effectively communicate, help reduce stress and workload, and support teams during critical care?

One significant insight came from the team’s first field deployment at Base Camp, a large-scale pediatric simulation event hosted by Weill Cornell Medicine that draws over 130 healthcare professionals across the U.S. and Canada.

“We found that the robot was very helpful in reducing mental, temporal, and physical demands,” Taylor said. “But then in terms of frustration, we did not see a reduction.”

That disconnect led the team to dig deeper into how healthcare workers understood the robot’s behavior, particularly when it opened drawers to suggest supplies. Even though participants had been briefed ahead of time, the stress of the scenario left many unsure of the reasons behind the robot’s decisions.

“The robot needed a very clear and direct way of communicating its intentions to the team,” said Taylor.

In response, the team added new ways for the robot to communicate. In subsequent prototypes, they added LED light strips as a visual cue of which drawer the robot was recommending for supply retrieval, and experimented with speech-based guidance to offer task reminders and supply locations. They also conducted lab-based studies with 84 participants to evaluate object search guidance and verbal task reminders.

A front and back view of a later prototype of the medical crash cart, alongside diagrams of its speech and drawer interfaces.
A prototype of the medical crash cart with drawer lights and speech-based guidance.

One discovery: positioning matters. Visual cues were only effective for people directly in front of the cart. Meanwhile, speech could reach more team members, but not always clearly.

“The environments are often very loud,” Taylor said. “So many times, the participants couldn’t even hear what the robot was saying.”

Long or vague spoken messages added to the challenge. Taylor noted that participants often became frustrated when the robot’s speech cues weren’t direct enough. In response, the team refined the robot’s spoken instructions to make them as concise and straightforward as possible.

Building on what they learned from these early deployments, the team is now exploring how robotic crash carts could support medical training and hospital-wide coordination. A multi-site study is currently underway to inform future iterations of the robot.

This project brought together a cross-institutional team of designers, engineers, and healthcare experts. Taylor’s team — including assistant professor Thijs Roumen of Cornell Tech and Cornell Bowers CIS; Ph.D. students Tauhid Tanjim and Michael Joseph Sack; research assistant Kexin Cheng; and Cornell Bowers CIS Associate Professor Malte F. Jung — collaborated closely with Dr. Kevin Ching and Dr. Jonathan St. George of Weill Cornell Medicine. The project also benefited from the contributions of Maia Hirsch, of the Mechanical Engineering Department at the Israel Institute of Technology, and Hee Rin Lee, of Michigan State University.

As healthcare systems face rising burnout and complexity, the team’s research offers a broader takeaway: thoughtful design can do more than assist in a crisis — it can reshape how teams communicate, coordinate, and care.

Sarah Marquart is a freelance writer for Cornell Tech.

< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES

By Sarah Marquart

Wireless health monitoring is rapidly evolving, promising a future where breathing, heart rate, and other key metrics can be tracked passively, continuously, and without the need for wearables, such as smartwatches, fitness trackers, or other bulky medical monitors.

However, with that promise comes a pressing concern: Who has access to the sensitive data collected by these devices, and how much control do users actually have?

That’s the question driving VitalHide, a novel privacy-preserving system developed by researchers at Cornell Tech. Assistant professors Rajalakshmi Nandakumar and Thijs Roumen, along with Cornell Tech Ph.D. students Zekun Chang, Yixuan Gao, and Tanvir Ahmed, unveiled the technology in February 2025 at the HotMobile conference in their paper, “VitalHide: Enabling Privacy-Aware Wireless Sensing of Vital Signs.”

VitalHide aims to restore agency to users without rejecting the benefits of wireless sensing. Roumen, who specializes in digital fabrication and is also affiliated with the Cornell Ann S. Bowers College of Computing and Information Science (Cornell Bowers CIS), emphasized the growing threat of unauthorized monitoring.

“Your TV might collect your heartbeat to gather information on what series excites you, a router at work might inform your employer of your focus during work, and a fan in the waiting room at your insurance company might collect your breathing pattern,” Roumen explained. “In a hospital, it may be great to wirelessly monitor your vitals, and you’d be happy to share that data, but you may want to exclude others from capturing this without your consent.”

That’s where VitalHide stands out. Unlike traditional privacy tools that scramble, block, or limit access only after data is gathered, VitalHide tackles the problem at the source. It uses tiny vibration motors or shape-changing textiles to generate false motion signals — like a decoy heartbeat — to confuse unauthorized sensors.

“Our solution directly ‘encodes’ the physical phenomenon (movement induced by breathing or heartbeat) and only allows trusted actors to access the raw data by sharing the key for decoding,” Roumen explained.

“Authorized devices, knowing this key, can filter out the fake motion signals and accurately recover the true vital signs,” explained Nandakumar, who is also affiliated with the Jacobs Technion-Cornell Institute and Cornell Bowers CIS. “Unauthorized devices, without knowledge of the distribution parameters, cannot distinguish between the real and fake signals and are thus misled.”

The result is seamless, privacy-preserving protection that doesn’t degrade performance for approved health systems or require identifying every nearby device.

Built on a combination of soft robotics, wireless sensing, and functional embroidery, VitalHide’s components can be embedded into clothing or everyday textiles.

The team is working to shrink the system of tiny motors and moving parts — called the actuation system — that creates the protective “fake” motions on the body, making it easier to wear in garments like jackets or vests. They’re also strengthening the system against emerging threats, such as machine learning-powered attacks that attempt to reconstruct biometric signals.

According to the researchers, future improvements include testing across different sensing systems like Wi-Fi and acoustic, as well as creating theoretical guarantees (clear, math-based rules) to show that the system will reliably protect privacy and deliver accurate results.

“It is important to note that VitalHide is not anti-technology,” said Roumen. “Instead, it provides users with the ability to intentionally consent to those who access the vital data. As a marathon runner, I happily track my heart rate with my Garmin watch, but I would not want my heart rate to be tracked by my insurance company.”

The researchers are also conducting user-centered studies to ensure the technology is not only practical but also empowering in practice.

“As new sensing systems are built, there is increasing awareness of privacy concerns, and efforts like VitalHide represent early steps toward designing technologies that recognize these challenges from the beginning,” said Nandakumar. “Building privacy-aware systems is becoming integral to advancing wireless health technologies, not something separate or conflicting with their benefits.”

Ultimately, VitalHide may mark a significant turning point in how society balances innovation with individual rights.

“Without most people’s knowledge, technology has gotten incredibly powerful at detecting and distilling data about our vitals,” said Roumen. “The latest development of this being done wirelessly raises a red flag for how much we should blindly rely on such technological usage and where we draw the line for our privacy. Now is a good time to give people the ability to consciously decide who accesses their data and who doesn’t.”

Sarah Marquart is a freelance writer for Cornell Tech.

< Back to News

Media Highlights

Bloomberg Law

Ripple Ruling Blurs Definition of Cryptocurrencies as Securities

Mental Daily

Study Takes A Closer Look At NYPD Patrol Patterns Using Dashcam Footage

Tech Policy Press

Content Moderation, Encryption, and the Law

Princeton University

Tech Expert Arvind Narayanan Takes the Helm at Princeton Center for Information Technology Policy

Marktechpost

A New AI Research from Stanford, Cornell, and Oxford Introduces a Generative Model that Discovers Object Intrinsics from Just a Few Instances in a Single Image

RELATED STORIES