Raja Parikshat

Developing real-time, intelligent AI avatars, focusing on state-of-the-art single-image animation pipelines. My work involves enhancing lip-sync accuracy and visual quality, as well as integrating agentic AI capabilities to enable advanced, intelligent functionalities within the avatar system.

Worked on enhancing the lip-sync network to improve articulation and expressiveness of virtual avatars. Collaborated closely with the Performance Capture and Generative AI teams to develop large-scale, controllable avatar systems. For the V2 avatars, I led the end-to-end development of the lip-sync system, achieving results comparable to leading industry competitors. Additionally, in collaboration with the Performance Capture team, I contributed to the implementation of a real-time facial tracking solution.

Wrote an app in Python, to process data output by NLP models, to score LinkedIn profiles for recruitment. Tiktok Dynamic Filter generation using first-order motion model. Retrained it by reducing the size of the generator module and adding additional data.