My work spans computer vision, medical AI, natural language processing, large language models, and MLOps — the full path from research idea to deployed system.

Currently, my research focuses on building agentic AI systems capable of detecting, explaining, and mitigating model drift in medical imaging. Models degrade quietly after deployment; in healthcare, that quiet degradation matters most. I build the systems that notice.

Outside of research, I care about the engineering that makes models useful: reproducible pipelines, honest evaluation, calibrated confidence, and deployments that survive contact with real data.