Summary
Joanna Hong is a Research Scientist on the Speech Team at Google DeepMind with eight years of experience advancing speech and multimodal AI. She specializes in audio-visual speech modeling and human-centric interactions, applying large language model techniques to make speech-based AI more natural and robust. Her Ph.D. work at KAIST produced contributions in audiovisual speech recognition, lip-to-speech generation, and multimodal representation learning, and she has continued this trajectory through internships and research roles at Meta and Trillion Labs. Joanna was a founding technical team member at Trillion Labs contributing to a multilingual 7B LLM optimized for deployment, reflecting her blend of foundational research and production-aware model engineering. Based in New York, she combines deep academic rigor with hands-on system and training infrastructure experience, uniquely focusing on recovering or inferring missing modalities in real-world multimodal inputs. Colleagues describe her work as bridging lab-grade innovations and deployable speech solutions that improve human–AI conversational fluency.
8 years of coding experience
6 years of employment as a software developer
Ph.D. Candidate, Electrical Engineering, Ph.D. Candidate, Electrical Engineering at Korea Advanced Institute of Science and Technology
English, Korean