Link principal investigators in Earth Sciences, Geography, Mathematics, Astronomy, EEOB, Statistics, and other natural and mathematical science units prior to funding calls — enabling rapid, coordinated responses to short-lead-time opportunities. This is a deliberately symbiotic relationship: domain scientists gain access to state-of-the-art AI methods, while multi-modal observational datasets — from satellite archives to astronomical surveys — offer AI researchers novel, challenging problems far beyond standard benchmarks.

Upgrade OSU's Unity HPC cluster with modern multi-GPU nodes and up to 1 petabyte of managed, shared storage so research teams can access common datasets and reproducible code across projects. Coordinate with the Ohio Supercomputer Center (OSC) for burst capacity, technical support, and optimized storage access.

Employ a college-level data engineer to manage multi-sensor data ingestion, pre-processing, and fusion; establish best practices in version control, documentation, and reproducible AI workflows — so that code and datasets survive beyond individual grant cycles and trainee departures.

Fund an undergraduate manual-labeling pool to generate training data for supervised AI algorithms; host annual workshops, boot camps, and hackathons; offer faculty training modules in AI methods; and provide "red-team" proposal reviews in partnership with ERIK — all to lower the cost of adopting AI across the natural and mathematical sciences.

Launch a Joint Industry Consortium of member companies that receive early access to cutting-edge research through annual workshops and reports; collaborate with OSU's Technology Commercialization Office on IP protection and licensing; and partner with OSU Advancement to establish an Endowed Faculty Chair and graduate/postdoctoral fellowships in AI + remote sensing.

Coordinate AI, data-centric, and computational course offerings across units; develop an AI certificate program for students in the natural and mathematical sciences; and foster connections between research and the classroom — building a pipeline of AI-ready graduates for academia, government, and industry.

Pre-train self-supervised neural networks on petabyte-scale scientific datasets — from multi-sensor satellite archives to large astronomical surveys — then fine-tune (transfer learning) these foundation models for specific science objectives, dramatically reducing the labeled data and compute required for each downstream application.