Research

I study what breaks when AI leaves the lab—especially domain shift, compression, and generators the benchmark never saw. Full paper titles, methods, and numbers live on the Projects page.

Venues & recognition

IEEE ICAIC 2026 (Houston) — two papers accepted

Domain-shift protocol and Enhanced Model Paradox; quad-stream deepfake detection. Both patent pending.

MIT IEEE URTC 2025 — GAN-based satellite deepfake detection (IEEE Xplore forthcoming)

700,000+ real and synthetic satellite images; presented at ESRI IUC (recognized by Jack Dangermond); invited talk at King's College, University of Cambridge, before Nobel Laureate Thomas Cech.

Patents — 2 pending; Davidson Fellowship — submitted, pending

Patents cover domain-shift evaluation methodology and quad-stream detection architecture. Davidson submission on domain shift in deepfake detection.

Conrad Challenge — DeepTrust top 5 of 8,000+ teams (2025); top 100 of 24,000+ participants (2025)

Competition validation alongside peer-reviewed publication track.

Stanford Pre-Collegiate Summer Institute — Discrete Mathematics (2024); faculty recommendation

Intensive summer program in discrete math and proof-oriented work.

Publications & detail ->