Proposal

Narrative

Use of GenAI is continues to increase and preparing EECS students to effectively use GenAI is critical. This project requests budget to purchase LLM services for use by students. E.g., credits on OpenAI, Anthropic, Gemini/Google that students can use in their CS classes and capstone design projects. The use will be implemented in two upper division undergraduate classes and two graduate classes. Students in these classes will be able to use multiple LLMs and learn how to use these LLMs effectively. Students in these classes will receive LLM usage credit (e.g., up to $50 per student; average use is expected to be lower). To facilitate easy deployment and scaling of services, we plan to use the LLM gateway from TrussedAI to allow students to access any LLMs within their allocated budget. We have been using Trussed AI's LLM router since Summer 2024 and find it very useful. Trussed Gateway will allow is to manage student access and spend limiting. Use in Undergrad classes: ENG 4950C (enrollment 300 per year) and COP 4808 (enrollment 150 per year) Use in Grad classes: COT 6930 GenAI for Software Development Life Cycles (enrollment 250 per year)

Facilities

Hardware Requirements

no special hardware requirement

Software Requirements

needs a VM instance to run Trussed controller service. Trussed Controller license at $20K/year; $60K for 3 years

Personnel Costs

None, TAs and IT team will help with installation

Other Costs

Cost of LLM services used by students: $30K/year; $90K for three years This is the cost of accessing OpenAI, Gemini, LLAMA and other similar models

Timeline

Spring 2025 -- Purchase trussed controller license Summer 2025 -- pilot use in one undergraduate class Fall 2025 onward -- extend use to undergrad and grad classes

Sustainability

Will need budget to renew license beyond three years. May request additional techfee or request support from College depending on the use and the evolution of needs. We will explore the use of locally run LLM service such as LLAMA as an alternative to purchasing access to commercial LLMs.

Resource Matching

Department TAs will support the deployment of GenAI in classes.

Implementing Organization

Electrical Engineering and Computer Science with the help of our TSG group.