Developed the domain adaptation strategy for ambient medical AI, leveraging high-quality "golden" clinical data to fine-tune open-source foundation models. Implemented supporting multimodal (text and audio) infrastructure on GKE, integrating real-time automatic speech recognition and async audio processing.

(Acquired by Commure)

As a part of AWS AI team, I focused on enhancing the availability, distribution, and performance of Deep Neural Network models.

Advisor: Prof. Stefano Soatto