Kaizhuo Yan

Machine Learning Engineer Intern @TikTok

San Jose, CA, US

End-to-end trained a large-scale multimodal recommendation model.Built large-scale training datasets for LLM-based recommendation by cleaning raw interaction logs and constructing high-quality user–content pairings.(Also benchmark)Developed an encoder-based multimodal architecture to align short-form video features with recommendation embeddings, enabling LLMs to jointly reason over visual and system-level signals.Trained multimodal LLMs on TikTok user history to model preferences and generate personalizedrecommendations, incorporating both textual and visual modalities for improved predictive accuracy.Developed AI Agents powered by large multimodal models with tool-use capabilities.

University of Illinois Urbana-Champaign

Master of Science - MS

University of Illinois Urbana-Champaign

Bachelor of Science - BS