Research at the University is largely supported by grants from large funding organizations, but many grant applications are rejected because any number of strict application requirements are not met. Students on the U-M CAEN Grant team will create an AI supported tool to review grant submissions against the submission requirements.
Abstract:
Faculty receive most of their research funding from large funding organizations such as the NSF National Science Foundation, NIH National Institute of Health, and the US Department of Health and Human Services. Faculty typically submit proposals in response to a specific offer of funding for a particular area. In 2022 the University of Michigan spent $1.86 Billion dollars on research expenditures.
These proposals are central to securing funding and must adhere to both
- Content requirements by the funding body (e.g., innovating idea to solve X, specific budget for Y number of years, coordination or joint work with specific partners, etc.), and to
- Format and submission requirements (e.g., page limitations, formatting instructions, date submitted, etc.)
Proposals that make mistakes in any of the requirements are rejected and returned to the faculty. There is typically no opportunity for revisions or later submissions, and there are hundreds of ways a proposal could make mistakes.
The student team will create an AI-supported tool to automatically check grant submissions (uploaded as a pdf) against the requirements of both the grant and the granting body and return to the faculty a list of errors or likely errors to be investigated via a basic web interface. They will demonstrate their prototype on one class of grant applications.
The project will likely incorporate the techniques listed below, providing an outstanding opportunity to build skills. We expect that the solution will include:
- Web front end upload interface
- No login necessary
- Does not store history
- Auto-deploying with tagging
- Server-side compute for processing files
- Auto-deploying with tagging
When processing files, we would prefer an on-demand compute type (ie, AWS Lambda, AWS Fargate). This may not be practical once the stack is built out, so thought should go into the design so that it can be deployed to a static server (ie, EC2) if necessary. Since we are not storing past runs, a database may not be required.
The prototype will be developed using the following tech stack. We don’t expect students to know all parts of the tech stacks, but you must be motivated to build your skills and become proficient.
- (git (via GitHub)
- AWS
- Vue 3 / typescript
- OpenAI APIs
- LLM Framework (LangChain / LangSmith / LlamaIndex)
- Python
- Docker
- Vector store database (if required)
- MySQL or Postgres (if DB required)
Impact:
A grant tool would support an increase in the number of successful grants for UM faculty, which will result in more research awards for the University of Michigan. This money will support additional graduated students, have a societal impact through discoveries, and continue the reputation of Michigan as a leader and best. Additionally, this will reduce the time burden on faculty, who spend time on discounted grants based on technicality as opposed to the proposal’s merit.
Scope:
Minimum Viable Product Deliverable (Minimum level of success)
- Develop an understanding of the current business process, and technical challenges, interview important stakeholders, articulate current failure modes, and develop the most relevant use cases.
- Literature review of all relevant techniques for converting the PDF into something that can be evaluated, and if there are similar applications that review grants,
- Demonstrate functional competence in the tech stack by completing a “mini project”. Note – it is unlikely that any student would be fully competent in the entire tech stack before the project. Significant individual training effort is expected and required.
- Complete first prototype and demonstrate the functionality of the v1 prototype against applicable system requirements (before the end of the Winter term). Develop a strategic plan to address the highest priority development in a v2 prototype.
Expected Final Deliverable (Expected level of success)
- Complete the v2 prototype and demonstrate the results to the stakeholders.
Stretch Goal Opportunities: (High level of success)
- Provide suggested fixes (presumably it will only say what is wrong, not how to fix)
- Create a database of successful + unsuccessful grant applications, along with the pertinent criteria for each submission.
Below are the skills needed for this project. Students with the following relevant skills and interest, regardless of major, are encouraged to apply! This is a team based multidisciplinary project. Students on the team are not expected to have experience in all areas, but should be willing to learn and will be asked to perform a breadth of tasks throughout the two semester project.
AI Experience (4 Students)
Specific Skills: General knowledge and skills in Machine Learning/Artificial Intelligence, experience incorporating ML and AI techniques into general programming front end/back end.
EECS 281 (or equivalent) is required, Students should have completed at least 1 course in AI or Machine Learning. Please indicate this on your experience and interest form.
Likely Majors: CS, ROB, ECE, Data
General Coding (2 Students)
Specific Skills: General programming skills, good software engineering practice, and design
EECS 281 (or equivalent) is required, experience in full stack development a plus.
Likely Majors: CS, DATA, BBA
Human Systems Development (1 Students)
Specific Skills: Business process mapping, error identification.
Likely Majors: IOE, BBA, CS
Additional Desired Skills/Knowledge/Experience
Sponsor and Faculty Mentor
Nicole Heffernan
Director of Web Services at CAEN
Nikki, a U-M alumnus, has spent the last 24 years in IT starting as a web developer and now leads a team of developers, business analysts, and project managers that work on a variety of web and other IT projects for the college. Nikki enjoys working with others to find creative solutions to problems.
Weekly Meetings: During the winter 2025 semester, the U-M CAEN Grant team will meet on Tuesdays from 3 – 5 PM. Location TBD.
Work Location: The work will take place on the Ann Arbor campus.
Course Substitutions:CE MDE, ChE Elective, CS Capstone/MDE, Data Science MDE/Capstone, EE MDE, CoE Honors, SI Elective/Cognate
Citizenship Requirements: This project is open to all students on campus. International Students: CPT declaration (curricular practical training) is NOT required for this project because the sponsor (CAEN Department) is part of the University.
IP/NDA: Students will sign standard University of Michigan IP/NDA documents.
Summer Project Activities: No summer activity will take place on the project.
Learn more about the expectations for this type of MDP project
If you have any of these characteristics, highlight them on your Experience and Interest Form and talk about them in your (optional) one way video interview.