Projects

On-device small-language-model agents that autonomously control drones in contested environments — flying to objectives, identifying mission-relevant objects, and reprioritizing the mission as new observations arrive.

AI/ML lead on a $10M DIU effort for the Defense Health Agency, building biosurveillance capabilities focused on the early detection and prevention of pathogen outbreaks.

A method for reverse-engineering black-box AI models using adversarial AI.

Specialized RAG pipeline for writing proposals to contracts.

Transformer-based multimodal deepfake detection model, demoed to the DIU.

Two GitHub-based LLM agents for streamlining the software development lifecycle.

Managed an HPC compute environment used for Booz Allen's most compute-heavy on-prem workloads.

Python-based mathematical optimization models; expanded a warehouse optimization system across the Northeast region.

A series of weekly computational modeling projects covering ODEs, chaos, population models, optimization, and neural networks.