Research

My current research centers on AI and computer vision for quantitative phenotyping in neuroscience. I develop markerless pose estimation, behavioral-state modeling, and multimodal analysis pipelines that convert video into objective measurements of movement, pain-related behavior, and recovery after spinal cord injury.

View publications

Current Focus

AI for Behavioral Phenotyping

I use computer vision and deep learning to track and quantify animal behavior after spinal cord injury. Markerless pose estimation, combined with kinematic and state-based modeling, lets us measure locomotion, pain responses, and recovery from video — connecting what an animal does to what's happening in its nervous system.

Markerless pose estimation of freely behaving subjects
Kinematic profiling and behavioral-state segmentation
Automated quantification of gait, pain responses, and functional recovery
Phenotyping pipelines that combine pose, gait, and behavioral data

Try pose demo

Animated mouse locomotor skeleton rendered from PHRASE pose tracking coordinates on a dark background

Earlier Work

Viral Vector and Molecular Tool Development

Before pivoting to computational work, I built and tested adeno-associated viral vectors and CRISPR-based tools for neuroscience — constructs for labeling neurons, manipulating circuits, ribosomal tagging, and programmable RNA targeting, mostly applied to spinal cord injury models.

AAV reporter and ribosomal tagging constructs for neuronal labeling
RfxCas13d (CasRx) for programmable RNA targeting in the CNS
High-fidelity Cas13d variant comparison for reduced off-target activity
Optogenetic and chemogenetic tool delivery via AAV

Selected publications (8 total)

AAV8-hChR2-EYFP

Dual AAV co-injection showing green EFS-Cas13d neurons and magenta E2-Crimson-U6-DR30-gRNA expression

EFS-RfxCas13d-2A-EGFP + E2-Crimson-U6-DR30-gRNA