Junyi Chen (陈浚毅)

I am a researcher at ByteDance Global Monetization - GenAI. My work focuses on multimodal foundation models, video generation and editing, and large language models for recommendation. I am fortunate to be advised by Lu Chi, Jia Sun, Longteng Guo, and Zehuan Yuan.

Feel free to contact me through email.

Email / Google Scholar / GitHub

Research

Selected publications and projects.

	Bernini: Latent Semantic Planning for Video Diffusion 2026 VLM-guided semantic planning for controllable video generation and complex video editing. Paper / Project / Code
	HLLM-Creator: Hierarchical LLM-based Personalized Creative Generation 2025 Personalized creative generation with user modeling, data construction, and efficient inference for user-specific title generation. Paper / Code
	HLLM: Enhancing Sequential Recommendations via Hierarchical Large Language Models for Item and User Modeling 2024 A hierarchical LLM architecture for efficient long-sequence user interest representation in recommendation systems. Paper / Code
	EVE: Efficient Vision-Language Pre-training with Masked Prediction and Modality-Aware MoE 2023 Unified multimodal Transformer pre-training with modality-aware sparse Mixture-of-Experts and masked signal modeling. Paper / Code

Professional Experiences

ByteDance Global Monetization - GenAI
Researcher, Mar 2022 - Present, Shanghai

Worked on multimodal foundation models, video generation and editing, and LLM-based advertising delivery.

Built and optimized training for foundation video models, designed high-quality video editing data pipelines, and explored hierarchical LLM user modeling for recommendation and personalized creative generation.

Selected Awards

CCPC Changchun Site 2020, Silver Medal, Rank 30 / 264
ICPC Asia Shanghai Regional Contest 2020, Rank 83 / 675