Initial President's Innovation Fund awardees Lucas C. Parra (left) and Hernan Makse
Distinguished Professor of Physics Hernan Makse and Harold Shames Professor of Biomedical Engineering Lucas C. Parra are the initial recipients of the President’s Innovation Fund award at The City College of New York for their project, “Democratizing AI through Symmetry Compression: Scaling our current 10x Model Compression Technology to Commercially Relevant Large Language Models.”
Selected from a competitive field of 17 proposals spanning biomedical engineering, clean energy, AI and other disciplines, the pair’s project will receive $50,000 in funding over the next academic year to address issues concerning the rapid development of large language models, or LLMs, a type of artificial intelligence program that is trained on massive amounts of text to understand and summarize text, to answer questions, write code, do math and anything else that involved natural language.
The proposal focuses on three interconnected issues: developing a compressed model that would eliminate redundancies, essentially achieving the same performance at lower size and cost of the current models; massively reducing energy consumption by current LLMs; and overcoming the limitations of current compression methods, which save memory but sacrifice performance.
The end result would be to allow more competition for AI, as current models require capital and operational investments that are unaffordable for many enterprises.
“Every Google search you do runs a large language model, so there's billions of queries happening all the time to these models, and that consumes a large fraction of the cost,” said Parra, noting that as many as 400 trillion words are served by LLMs every day world-wide. “That's what takes about 90 percent of all the computations and, also, energy consumption.”
“That's where we come in,” he added. “We can take an existing trained model, and we can compact it by a factor of 10.”
“Our mathematical theory, drawn from the geometry of fibrations, identifies precisely which neurons of AI are duplicates and can be safely removed,” said Makse. “Our target is to achieve what nobody in the AI industry could: a hundredfold reduction in the size of frontier AI models, making them much cheaper to run. The math tells us that this is within reach. We are compressing intelligence so that AI can finally become universal.”
Assistant Professor of Computer Science Arthur Paul Pedersen, in collaboration with Associate Professor of Computer Science Michael Grossberg, Assistant Professor of Computer Science Saptarashmi Bandyopadhyay, and doctoral student in electrical engineering William Seong, received more than $8,000 in summer funding for the second place project, “FEAT (Forensic Expert Activity Assistance Tool),” which helps analysts evaluate competing explanations for DNA evidence using peer-reviewed scientific literature while preserving expert judgment at every stage.
“The goal is not to replace forensic expertise, but to make the reasoning behind forensic conclusions more transparent, reproducible, and defensible in court,” said Pedersen. “In many criminal cases involving DNA evidence, the central question is not whose DNA was found, but how it got there. Evaluating competing explanations for its presence — known as activity-level interpretation — requires forensic scientists to weigh scientific evidence, published research, and alternative scenarios, yet no dedicated software exists to support this work in a transparent and auditable way, underscoring the need for rigorously evaluated tools that make the underlying research, assumptions, and limitations accessible for scrutiny.”
PIF provides seed funding to faculty pursuing translational research with commercialization potential as they move commercially promising discoveries out of the lab and toward real-world impact. The application process involves submission of such an idea, process or concept to a select committee of senior CCNY academics, which forwards its recommendations to President Vince Boudreau for the final award decision.
“The PIF committee was thrilled to see such an exceptional pool of applicants in the inaugural PIF cycle, and we are so proud to support the winning proposals,” said City Innovations Collaborative Director Andoni Mourdoukoutas. “There's clear interest from CCNY faculty in commercializing their research, and our goal is to help support them on their journey through programming like PIF."
“The President’s Innovation Fund adds an important new layer to City College’s research apparatus and ambitions,” said President Boudreau. “So much of the research taking place in our laboratories and studios has in mind a public purpose, some innovation that will serve the interests of humanity. The President’s Innovation Fund sets out to support original research at the College with important translational potentials, so that something germinated in a City College lab will someday find expression as a realized innovation in our society.”
Applications for the 2027 PIF cycle are expected to be opened in February 2027. Eligibility, instructions and additional details are available at /presidentsoffice/presidents-innovation-fund
About the City Innovations Collaborative
The City Innovations Collaborative (CInC) supports CCNY faculty researchers in bridging the gap between discovery and real-world impact through programs, funding, mentorship, and connections to the broader innovation ecosystem. Its flagship program, the President's Innovation Fund (PIF), is a competitive internal grant that provides seed funding to faculty pursuing translational research with commercialization potential, and its strong inaugural cycle demonstrated significant latent demand among CCNY faculty for structured commercialization support. PIF is the initial kernel of a wider suite of complementary innovation programs CInC is developing over the coming year, including a flagship annual Innovation Day showcasing faculty projects to the NYC innovation community and a Technology & Innovation Subcommittee connecting faculty with industry partners and investors. Together, these programs are designed to meet faculty researchers at different stages of the commercialization journey, from early ideation through active venture development, positioning CInC as CCNY's primary interface with the broader New York City innovation ecosystem.
About The City College of New York
Since 1847, The City College of New York has provided a high-quality and affordable education to generations of New Yorkers in a wide variety of disciplines. CCNY embraces its position at the forefront of social change. It is ranked #1 by the Harvard-based Opportunity Insights out of 369 selective public colleges in the United States on the overall mobility index. This measure reflects both access and outcomes, representing the likelihood that a student at CCNY can move up two or more income quintiles. Education research organization Degree Choices ranks CCNY #1 nationally among universities for economic return on investment. In addition, the Center for World University Rankings places CCNY in the top 2.2% of universities worldwide in terms of academic excellence. Labor analytics firm Lightcast puts at $3.7 billion CCNY’s annual economic impact on the regional economy (5 boroughs and 5 adjacent counties) and quantifies the “for dollar” return on investment to students, taxpayers, and society. At City College, more than 16,500 students pursue undergraduate and graduate degrees in eight schools and divisions, driven by significant funded research, creativity, and scholarship. In 2023, CCNY launched its most expansive fundraising campaign ever. The campaign, titled “Doing Remarkable Things Together,” seeks to bring the College’s Foundation to more than $1 billion in total assets in support of the College's mission. By May 2026, more than $500 million had been raised. CCNY is as diverse, dynamic, and visionary as New York City itself. View CCNY Media Kit.
Syd Steinhardt
212-650-7875
ssteinhardt@ccny.cuny.edu