Ideas

An AI-powered course scheduling system for University of Michigan students that integrates with Atlas data. Students can input their academic plan, ideal workload, and preferred class times. The AI will analyze course availability, prerequisites, credit requirements, and create optimized schedules that ensure students don't miss any required credits for graduation. This would revolutionize the course registration and "backpacking" process, making it more convenient for students while also assisting academic advisors in their planning.

AI/ML · Education · Web Development

A real-time geomagnetic storm alert system built on top of the ML classifier I'm co-authoring at the Magnetometer Lab. The dashboard would ingest live Sym-H indices from NOAA/SuperMAG, run the Random Forest model in the browser or via a lightweight API, and surface storm vs. impulse probabilities with confidence intervals. The goal is to make the research immediately useful, giving satellite operators, power grid engineers, and space weather researchers a fast early warning signal that doesn't depend on solar wind data. A public-facing version could also be embedded in educational tools.

Space Physics · AI/ML · Data Visualization

A browser-based simulation tool for rapidly prototyping eVTOL flight envelopes during the early design phase. Given a set of motor configs, rotor geometries, battery specs, and payload constraints, the tool computes hover endurance, maximum range, climb rate, and dynamic stability margins. Targeted at student teams like U-M Vertical Flight who are iterating on airframe concepts without access to expensive CFD suites. Built in TypeScript with Three.js for 3D visualization and a physics engine for real-time parameter sweeps.

Aerospace · Simulation · Web Development

A bigger version of GoFundMI: a platform any university could use to connect student researchers with donors who care about their work. My thesis is that most research funding friction is really an information problem. Donors don't know which projects exist, and students don't know which donors are looking. A structured matching layer with research profiles, project milestones, and verified outcomes could open up a new kind of direct research philanthropy that skips the overhead of traditional grant pipelines.

Non-Profit · Web Development · Education