AI Research Transforms Industries
SHERI HALL
Over the past decade, discoveries in artificial intelligence and machine learning have demonstrated the potential to revolutionize virtually every sector of the global economy. As a result, the path from cutting-edge computer science research to groundbreaking business ventures has never been more direct.
At Carnegie Mellon University’s School of Computer Science, faculty and students are developing new technologies that are reshaping education, computer science, engineering and countless other industries. Their discoveries stimulate economic growth in Pittsburgh and across the globe.
NoRILLA, a mixed-reality learning platform that guides preschool and elementary students through interactive science experiments, demonstrates how augmented reality research can address educational challenges. Moonshot AI’s chatbot Kimi, co-created by computer science graduates Zhilin Yang (SCS 2019) and Yuxin Wu (SCS 2016), is making generative AI accessible to millions of Chinese-speaking users. And All Hands AI, founded by Graham Neubig, associate professor in the Language Technologies Institute, is revolutionizing software development by using generative AI to write computer code.
These are just a few of the many success stories that demonstrate how CMU researchers are translating technological breakthroughs into successful business ventures. They all share common threads: input from world-class SCS faculty members, access to cutting-edge resources and collaboration with industry partners.
Niloy Gupta (SCS 2015)
Niloy Gupta (SCS 2015) is an expert in machine learning who has been a part of four different technology companies since he graduated from Carnegie Mellon. He is currently a lead ML/AI engineer at Attentive, which offers AI-based solutions for digital marketing.
“I owe my career to the time I spent at CMU,” he said. “My work there profoundly shaped my understanding of what’s possible. We are just starting to scratch the surface of how AI can make our lives better.
“But more than that, the research that comes out of CMU and other top institutions is the ground that industry builds their castles on,” he continued. “Universities are the breeding ground for talent and innovation.”
Government Grants Plant the Seeds of Innovation
Funding from government agencies, particularly the National Science Foundation (NSF) and the Department of Defense (DOD), is an essential part of creating this breeding ground, Gupta explained.
That’s because investments in basic research allow faculty and students to pursue ambitious projects without the immediate pressure of commercial viability. When researchers have the freedom to explore new approaches, they can achieve breakthroughs that serve as the foundation for successful startups.
Graham Neubig, Associate Professor in the
LTI and Founder of All Hands AI
Graham Neubig, an associate professor of computer science in the LTI and founder and chief scientist of All Hands AI, started working on the foundational research for his company thanks to a grant from the NSF. All Hands AI uses software development agents to manage many of the tasks involved in software development such as writing and executing code, integrating with outside platforms and fixing bugs.
“We wanted to learn how to generate computer code from English prompts across a variety of software projects,” Neubig explained. “Now in 2025, everybody knows that’s possible. But when we started in 2017, people thought, ‘How could that be real?’
“The foundational NSF grant was essential,” he continued. “The important thing to understand is that we received funding for the project before we knew it was feasible. Government funding is one of the only ways to start something brand new.”
Today, All Hands AI offers its groundbreaking coding agent to anyone who wants to use it.
“AI really lifts everybody up a level,” Neubig said. “People who wouldn’t be able to build a website can now use our tool to build a website. People who know some coding can build an even bigger or better piece of software. And software developers can do even bigger things, or take on larger projects because AI saves them time.”
All Hands AI is working to make their open-sourced, web-based tool as useful as possible for individual developers. In addition, they offer a commercial tool that integrates with other apps, such as GitHub and Slack. And they provide enhanced security and multiuser authentication to enterprises looking for more features.
Meanwhile, researchers in Neubig’s lab currently receive funding from the NSF and the Air Force Research Laboratory to continue working on machine translation and multilingual natural language processing.
“My ultimate goal is that every person in the world should be able to communicate with each other and with computers in their own language,” he said. “I’m excited about building a community around entrepreneurship within SCS.”
From Innovative Ideas to Real-World Solutions
In fact, SCS is ideally positioned to develop new businesses, said Eric Xing, professor in the Machine Learning Department with appointments in the LTI and Computer Science Department.
“CMU takes a unique approach to AI production and innovation, which is to focus on the engineering details and elegant solutions to complex problems,” he said. “Many CMU graduates staff the top technology companies in the world. CMU has a unique role in the global space.”
Xing’s research on machine learning, statistical methodology and large-scale computational systems is breaking new ground. His teams are developing biological AI models that can analyze genomes and create frameworks for complex biological systems. And they are working to develop artificial general intelligence (AGI) models that understand, learn and apply knowledge across a wide range of tasks.
“I want to build AI models that are smart in planning and reasoning operations for real-world environments — how to run a company, for example, or how to build interesting games,” said Xing. “It’s not just reasoning to show off fancy capabilities, but actually to accomplish missions.”
Since joining CMU in 2004, Xing has spun off two successful companies.
In 2016, he founded Petuum Inc., whose goal was to help organizations adopt machine learning and generative AI platforms to support their business models. “That company was in many ways ahead of its time,” Xing said. “We were trying to build the software infrastructure for producing large-scale AI systems years before those systems were broadly available. The fundamental technology for building large language models originated from Petuum research and prototypes.”
Last year, Xing founded GenBio AI, where he currently serves as chief scientist.
Eric Xing, Professor in MLD, LTI and CSD,
and Founder of Petuum Inc. and GenBio AI
In addition to these ventures, Xing founded the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) in 2019, a graduate research university in the United Arab Emirates dedicated to advancing AI as a global force.
In the past 20 years, Xing and his students have received millions of dollars in grants from federal agencies including the NSF and the National Institutes of Health. Xing credits the U.S. system of government-funded public research with building entrepreneurial strength in America.
“Federal funding generates knowledge and ideas, not just intellectual property,” he said. “This research can be taken further to create products, but it also leads to even more new ideas. And working on these projects just makes scientists smarter. This is different from the business world, where companies invest money and measure the outcome. Government money is meant to do what industry cannot do in terms of promoting research and sharing that knowledge in the public space. This translates into building the economy.”
A Culture of Excellence Through Entrepreneurship
The translation that Xing speaks of is something Gupta has witnessed many times over, having worked at large, established tech companies and startups.
After CMU, Gupta first got a job as a machine learning engineer at Yelp, where he set himself apart by using machine learning to scale advertising algorithms. He was then recruited as the first machine learning engineer at Affirm, a financial technology company that allows consumers to finance purchases with flexible payment options. There, he led teams building machine learning systems that could optimize loan terms, interest rates and repayment behaviors.
Gupta went on to serve as a co-founder of Lambent Logic, which developed big data solutions to help pharmaceutical companies navigate the distribution process, manage revenues, identify revenue leakages and simplify their government pricing structures.
And last year, he joined Attentive Mobile as a lead machine learning engineer. The technology company uses AI and machine learning to customize direct-to-consumer marketing campaigns for more than 8,000 leading brands such as Neiman Marcus, Samsung, Wayfair and Dyson.
“My professional success is based on the knowledge gained at Carnegie Mellon,” he said. “It’s resulted in hundreds of millions of dollars in added value to the economy. But more than that, the culture at CMU — of honesty and commitment to excellence — cultivates professional pride that carries through your career.” ■
GenBio AI
Eric Xing considers GenBio AI to be, at its core, an AI company, not a healthcare, medical or biotech company, that has learned from large language models’ (LLMs) broad purposes.
“We are creating an AI system based on biological and medical data so you can use it to answer complicated medical questions, such as what type of treatments pharmaceutical researchers should pursue,” Xing said.
ChatGPT and other LLMs can be used widely to answer a variety of language and conversational questions. In the same way, GenBio uses generative AI technologies to create a multiscale biological model — which the company calls an AI-driven digital organism (AIDO) — that can predict, simulate and program biology at all levels. Researchers run their ideas through the AIDO before they do trials in a wet lab. This helps identify promising solutions and narrow down less plausible ones before spending the time and money to set up experiments. It can also reduce the risk involved in handling viruses and other materials in labs. That is a departure from the way drug development has traditionally been conducted, focusing narrowly on one problem or disease and relying on data from specific populations.
“What GenBio AI is trying to do by building a foundation model for AIDO is to disrupt the whole workflow of biomedicine, meaning that starting from target discovery all the way down to protein or vaccine design, then trial, then to toxicity study, then to personalization and repurposing — you can use this as one system to address all these problems, which are actually the same problem at different scales,” Xing said.
This year the U.S. Food and Drug Administration published new recommendations for using AI in drug development that align closely with GenBio AI’s vision, encouraging scientists to start the drug development process by using AI before going to a wet lab.
GenBio AI began as a global company with offices in Silicon Valley, Paris and Abu Dhabi. Ziv Bar-Joseph, the FORE Systems Professor of Computer Science at SCS, left his role as head of R&D Data and Computational Sciences at Sanofi to join GenBio AI as co-founder. ■
