Research

I am open to research collaborations. Feel free to drop me an email!

Projects

nxtAIM 2024–present

A federally funded research initiative using generative AI to create photorealistic synthetic driving data for training and validating autonomous driving perception. Leading Data Consolidation and Foundation Model training across consortium partners.

[project]

Publications

Localized Semantic Alignment for Temporally Consistent Driving Video Generation Published as Oral at IV2026, 2026 · [arXiv] [code]

Overview of the LSA framework for improving temporal consistency in video generation

Overview of LSA: a semantic feature consistency loss aligns DINOv2 representations of generated and ground-truth frames within dynamic-object regions during training, while inference remains unchanged.