CSE DSI Research Support

Fall 2026 CSE DSI Seed Grant RFP

(Deadline: Monday, November 9, 2026)

Purpose

The CSE Data Science Initiative (DSI) invites proposals for seed funding focusing on AI for Science and the Science of AI.

Building on the discussions and connections developed through our recent AI for Science workshops, this round has a particular goal: to catalyze strong, multidisciplinary UMN teams that can pursue major emerging opportunities in AI for Science and the Science of AI.

Rapidly growing federal investment in these areas is creating significant new opportunities for research at the intersection of AI, science, and engineering. These include priorities identified in the White House AI Action Plan (opens pdf), the Genesis Mission (opens pdf), Science: A New Golden Age (opens pdf), and emerging programs across federal agencies (e.g., the National Security Science & Technology Strategy (opens pdf)).

CSE is particularly well positioned to respond to these opportunities, with expertise spanning foundational AI and data science as well as a broad range of scientific and engineering disciplines. This seed program is intended to bring the existing UMN strengths together, develop promising new research directions and preliminary results, and position UMN teams for significant external funding.

A New Emphasis on Building Larger Teams

This round will place particular emphasis on larger, well-integrated teams that bring together complementary expertise across AI, science, and engineering.

We are especially interested in teams that can address ambitious scientific problems while also identifying opportunities for advances in AI itself. Projects may therefore contribute to AI for Science, the Science of AI, or both.

Teams do not need to be fully formed before participating in the program. The two fall team-building meetings described below are intended to help faculty identify opportunities, find collaborators, and develop research concepts.

Particularly compelling larger teams may be considered for more than one graduate research assistantship.

Team-Building Meetings

CSE DSI will host two meetings this fall to facilitate collaboration and the development of competitive teams.

September 18: Opportunities and Team Formation

Friday, September 18, PAN 110, 10 am-noon

Register for these meetings by Tuesday, September 8.

The first meeting will include:

A brief overview of emerging federal funding priorities (opens a Google doc) and opportunities in AI for Science and the Science of AI
Key research themes and opportunities emerging from our February 2026 GenAI4Science workshop and August 2026 Frontiers of GenAI & Science workshop
Lightning talks from participating affiliates/teams proposing new research directions and team formation
Networking and opportunities to identify potential collaborators and form teams

Affiliates are welcome to participate regardless of whether they already have a research concept or partial team, or are primarily interested in exploring opportunities and finding collaborators.

October 9: Developing Teams and Research Concepts

Friday, October 9, Keller 3-180, 1-4 pm

The second meeting will provide additional opportunities for team formation and development. Emerging teams will have an opportunity to discuss promising research directions, present developing concepts, and receive feedback from colleagues before preparing their seed proposals.

Research Scope and Priorities

Proposals may address a broad range of compelling research challenges where advances at the intersection of AI, science, and engineering can make a major difference. We particularly encourage projects in which collaboration across disciplines is integral to the research rather than simply an application of an existing AI method to a new dataset.

Projects may, for example:

Develop new AI methods motivated by challenges arising in scientific and engineering domains
Incorporate scientific knowledge, structure, constraints, or reasoning into AI models
Advance reliable, interpretable, and scientifically valid AI methods
Develop generative, foundation, multimodal, or agentic AI approaches for scientific discovery
Address important scientific and engineering challenges in areas such as materials, manufacturing, physical sciences, climate and environment, health, agriculture, and other areas of societal importance
Develop datasets, benchmarks, evaluation frameworks, or testbeds that enable broader advances in AI for Science

These examples are illustrative rather than exhaustive. We welcome other ambitious directions that can position UMN teams for leadership in emerging AI for Science opportunities.

Team Expectations

Proposals should bring together complementary expertise from AI/data science and relevant scientific or engineering disciplines.

While proposals involving faculty from at least two departments are eligible, priority will be given to larger, well-integrated multidisciplinary teams when the scope of the research and the external opportunity justify such a team.

Proposals should clearly explain:

Why the research problem requires the proposed combination of expertise
The distinct roles and contributions of team members
How the collaboration creates capabilities or research directions that would be difficult for individual investigators to pursue independently

External Funding and Growth Potential

A central objective of this program is to help teams become competitive for substantial external funding.

Proposals should identify one or more relevant external opportunities or sponsors and explain how the seed project will position the team to pursue them. These may include opportunities from NSF, DOE, NASA, NIH, USDA, DoD, other federal agencies, industry, or foundations.

Alignment with major national initiatives such as the White House AI Action Plan (opens pdf), the Genesis Mission (opens pdf), and other emerging federal AI for Science priorities is encouraged where appropriate, but proposals are not limited to these programs or topics.

Funding and Support

CSE DSI expects to support up to six graduate research assistantships across the selected projects.

Each graduate research assistantship will provide support for one year (20 hours/week), beginning in Spring 2027. Graduate assistants must be CSE students and should be jointly advised by members of the multidisciplinary team.

Depending on the strength, size, and scope of the proposed effort, a team may be considered for more than one graduate research assistantship.

The purpose of this support is to enable teams to establish their collaboration, develop preliminary results, and prepare for significant external proposals.

Proposal Format

Proposals should be submitted as a PDF and should not exceed three pages (single space, font size 11 or follow the NSF guidelines) including figures, tables, and references. Proposals requesting more than one graduate research assistantship can add an extra page, and submit up to four pages in total.

Each proposal should address:

Research opportunity: What important scientific, engineering, and/or AI challenge will the team address? Why is this opportunity timely and significant?
Research innovation: What is the relevant state of the art and its limitations? What is new about the proposed approach, and why is it likely to succeed?
Research plan: What are the key research questions or hypotheses, proposed activities, expected outcomes, and milestones? Identify relevant datasets, computational resources, experimental capabilities, or other resources as appropriate.
Team and integration: What expertise is needed to pursue the proposed research? What does each team member contribute, how will these complementary capabilities be integrated, and what does the collaboration enable that would be difficult for individual investigators to accomplish independently?
Use of seed support: How will the requested graduate research assistantship(s) be used to establish the collaboration and produce the preliminary results, methods, datasets, benchmarks, or other capabilities needed for the larger effort?
Path to external funding: What significant external funding opportunity or opportunities could grow from this work? If specific funding programs are known, identify them and explain what the team needs to accomplish during the seed period to become competitive.

Deadline: Monday, November 9, 2026
Submission: Send proposals to [email protected]

Selection Priorities and Review Criteria

Proposals will be evaluated based on the extent to which they:

Bring together a strong, well-integrated multidisciplinary team, combining complementary expertise in AI/data science and relevant scientific or engineering domains.
Address an important research opportunity at the intersection of AI and science or engineering, with the potential to advance scientific discovery, AI methods, or both.
Pursue an innovative research direction in which the integration of AI and domain expertise is central to the proposed work.
Use the seed funding strategically to build or strengthen a multidisciplinary collaboration and develop preliminary results, datasets, benchmarks, methods, or other capabilities needed for a larger research effort.
Demonstrate strong potential for external growth, including a credible path toward significant funding from federal agencies, industry, or foundations and, where appropriate, alignment with emerging national AI for Science priorities.

Eligibility

Proposals must include at least one CSE faculty member. Graduate students supported through this seed program must be enrolled in a CSE graduate program.

The CSE DSI is able to provide seed funding for small projects at the initial stage of research that bring college faculty in data science and science and engineering together.

This funding is in the form of graduate student assistantships (formerly fellowships) where the student is participating in the research and is mentored by at least two faculty from different departments. The initial results from these small projects are expected to form the basis for future larger research proposals.

The call for proposals is usually announced as part of our community-building exercises around a specific topic area chosen in response to interests expressed by our affiliated faculty. The announcement will be circulated to all CSE DSI affiliates.

The table below shows the range of projects supported to date.

Research Topic	Graduate Assistants	Faculty
Accelerating materials discovery with adaptively guided Generative AI	Zichen Huang (CEMS) Jigyasa Gaurav (CS&E)	Chris Bartel (CEMS) Aryan Deshwal (CS&E)
Graph GenAI for Interpretable Combustion Reaction Mechanism Discovery in Propulsion and Energy Systems	Sitong Pan (CS&E) Pan Hao (CS&E) Haotian (James) Du (ME)	Qianwen Wang (CS&E) Suo Yang (ME)
GenAI-Enabled Representation of Quantum ChromoDynamics	Sitong Pan (CS&E) Peiran Li (PHYS)	Qianwen Wang (CS&E) Zhen Liu (PHYS) Nadja Strobbe (PHYS)
AI Data Scientists: Agentic Transformation from Data into Scientific Insights	Shuyu Gan (CS&E) Pan Hao (CS&E)	Dongyeop Kang (CS&E) Mingyi Hong (ECE) Qianwen Wang (CS&E)
Advancing Materials Discovery: Emulation of Molecular Ensembles and Dynamics	Youcheng Wang (CEMS)	Aryan Deshawn (CS&E) Sapna Sarupria (CHEM)
Physics-informed Generative Model for High-Resolution Traffic State Estimation and Optimal Sensor Placement	Yuxuan Zhao (MATH) Lindong Liu (CEGE)	Seongjin Choi (CEGE) Yulong Lu (MATH)

Research Topic	Fellows	Faculty
Towards Unconstrained Multi-sensor “State-of-the-Heart” Monitoring and Disease Prediction	Xiangzhen (Raven) Kong Yijun Lin	Alena Talkachova (BME) Yao-Yi Chiang (CS&E)
Large Scale Data Extraction from Population Sampling of Dispersed Waterborne Photovoltaic Microparticles	Tianqi Luo Yijun Lin	Joey Talghader (ECE) Yao-Yi Chiang (CS&E) Daniel Bond (Microbiology)
Can Physically Informed Deep Generative Models Improve Seasonal Predictability of Global Precipitation?	Reyhaneh Rahimi	Ardeshir Ebtehaj (CEGE) Vipin Kumar (CS&E)
Sharp Analysis of Atomic-Resolution STEM Data via Deep Learning	Zhong Zhuang Hengkang Wang	Ju Sun (CS&E) Andre Mkhoyan (CEMS)
Constrained Deep Learning for the Efficient Discovery of Stable Solid-State Materials	Jane Schlesinger Ryan Devera	Chris Bartel (CEMS) Ju Sun (CS&E)
Accelerating Novel 2D Material Discovery Through Machine Learning	Wei Ren Kyle Noordhoek	Ke Wang (Physics) Chris Bartel (CEMS)
Learning Using Privileged Information (LUPI) for Materials Discovery	Eng Hock Lee	Vladimir Cherkassky (ECE) Tony Low (ECE) Ellad Tadmor (AEM) Eric Fuemmeler (AEM)
Integrated Molecular Simulations and Machine Learning Tools to Uncover the Treasure Trove of Hidden Structures during Crystallization	Steven Hall	Sapna Sarupria (CHEM) Michael Steinbach (CS&E)
Computationally Efficient Bayesian Inference for Population Properties of Astrophysical	Xiao-Xiao Kou	Galin Jones (STAT) Vuk Mandic (PHYS)
Accelerating Electromagnetic Design and Simulations with Physics-Informed Artificial Intelligence	Binyao Guo Samuel Dietterich	Qizhi He (CEGE) Shaul Hanany (PHYS)
Event Identification from the Sun throughout the Solar System	JangHyeon Lee William Setterberg	Yao-Yi CHiang (CSE) Lindsay Glesener (PHYS)
Post-processing Techniques in Neural Gradient Networks with Applications to Inverse Problems in Astrophysics	Andrew Toivonen Akshay Kumar	Michael Coughlin Jarvis Haupt Vuk Mandic
Building Multi-modal Foundation Models for Supernovae Data	Felipe Fontinele Nunes Wenya Xie	Ray Liu Michael Coughlin

The CSE DSI is able to provide seed funding for small projects at the initial stage of research that bring college faculty in data science and science and engineering together.

This funding is in the form of graduate student fellowships where the student is participating in the research and is mentored by at least two faculty from different departments. The initial results from these small projects are expected to form the basis for future larger research proposals.

The table below shows the range of projects supported to date.

Research Topic	Fellows	Faculty
Towards Unconstrained Multi-sensor “State-of-the-Heart” Monitoring and Disease Prediction	Xiangzhen (Raven) Kong Yijun Lin	Alena Talkachova (BME) Yao-Yi Chiang (CS&E)
Large Scale Data Extraction from Population Sampling of Dispersed Waterborne Photovoltaic Microparticles	Tianqi Luo Yijun Lin	Joey Talghader (ECE) Yao-Yi Chiang (CS&E) Daniel Bond (Microbiology)
Can Physically Informed Deep Generative Models Improve Seasonal Predictability of Global Precipitation?	Reyhaneh Rahimi	Ardeshir Ebtehaj (CEGE) Vipin Kumar (CS&E)
Sharp Analysis of Atomic-Resolution STEM Data via Deep Learning	Zhong Zhuang Hengkang Wang	Ju Sun (CS&E) Andre Mkhoyan (CEMS)
Constrained Deep Learning for the Efficient Discovery of Stable Solid-State Materials	Jane Schlesinger Ryan Devera	Chris Bartel (CEMS) Ju Sun (CS&E)
Accelerating Novel 2D Material Discovery Through Machine Learning	Wei Ren Kyle Noordhoek	Ke Wang (Physics) Chris Bartel (CEMS)
Learning Using Privileged Information (LUPI) for Materials Discovery	Eng Hock Lee	Vladimir Cherkassky (ECE) Tony Low (ECE) Ellad Tadmor (AEM) Eric Fuemmeler (AEM)
Integrated Molecular Simulations and Machine Learning Tools to Uncover the Treasure Trove of Hidden Structures during Crystallization	Steven Hall	Sapna Sarupria (CHEM) Michael Steinbach (CS&E)
Computationally Efficient Bayesian Inference for Population Properties of Astrophysical	Xiao-Xiao Kou	Galin Jones (STAT) Vuk Mandic (PHYS)
Accelerating Electromagnetic Design and Simulations with Physics-Informed Artificial Intelligence	Binyao Guo Samuel Dietterich	Qizhi He (CEGE) Shaul Hanany (PHYS)
Event Identification from the Sun throughout the Solar System	JangHyeon Lee William Setterberg	Yao-Yi CHiang (CSE) Lindsay Glesener (PHYS)
Post-processing Techniques in Neural Gradient Networks with Applications to Inverse Problems in Astrophysics	Andrew Toivonen Akshay Kumar	Michael Coughlin Jarvis Haupt Vuk Mandic
Building Multi-modal Foundation Models for Supernovae Data	Felipe Fontinele Nunes Wenya Xie	Ray Liu Michael Coughlin