Led the one-year evaluation of the DoD's first Joint Cyber Command and Control platform. Gathered field requirements from Air Force, Army, and Navy operators, translated them into test architecture, identified 6 critical deficiencies, and tracked resolution across service stakeholders.

Deployed production GPT-4 pipelines that automated USAF operational reporting, cutting 800+ hours of manual work per year. Built data ingestion from structured operational feeds, multi-step agent workflows for summarization and decision support, and a Streamlit front end for non-technical operators.

GPU-accelerated CV system using YOLOv8x with SAHI sliced inference, GStreamer RTSP ingestion, and BotSort tracking. Processes aerial ISR video in real time to detect airfield damage. Containerized with a FastAPI test harness for repeatable regression testing across single-image, RTSP video, and batch modes.

Managed a 34-person multidisciplinary engineering team on a directed energy research program at AFRL. Designed a scalable pulsed laser system exceeding energy targets by 50% through optimized thermal management. Prevented a 6-month schedule slip through early risk-reduction analysis.

Built a synthetic imagery pipeline using Python, TensorFlow, and Blender to train a UNet segmentation model for automated aircraft surface inspection. Eliminated the need for expensive real-world annotation by training entirely on procedurally generated images. Reduced real-world data requirements by 90%.

Building production C#/.NET applications for DoD clients across the full stack: backend service architecture, REST API integration, and CI/CD-gated testing. Containerized deployments with Docker.