Role: Research Scientist, Meta Smart Glasses ยท July 2024โJune 2026
I worked on multimodal large language model systems for extended, context-aware humanโAI conversation across server and on-device experiences.
Selected contributions
- Led a global rollout that increased feature engagement by 14%.
- Developed text-and-audio reasoning approaches that increased engagement by 5% and reduced refusals by 4%.
- Collaborated across research and engineering teams to connect model quality, latency, and product impact through high-impact A/B testing.
Research focus
This work connected audio and language representations with practical deployment constraints, with emphasis on reliable evaluation, efficient inference, and natural interaction in wearable computing.