Research Areas

My recent work focuses on multimodal generation, large language models, AI agents, and recommendation systems. I am especially interested in turning foundation models into reliable production systems for advertising, content understanding, and creative generation.

Multimodal Generation

• Video and image generation with better controllability, consistency, and production quality.
• Efficient training and inference for large-scale creative generation systems.
• Editing and generation pipelines for advertisement videos and product-driven content.

Large Language Models and AI Agents

• Domain-adapted LLM systems for advertising, customer service, diagnosis, and workflow automation.
• Agentic systems that combine planning, retrieval, tool use, and evaluation.
• Long-context understanding, structured outputs, and real-world deployment for business scenarios.

Recommendation and Advertising Intelligence

• Generative recommendation for large-scale advertising platforms.
• Multimodal understanding of products, landing pages, and creative materials.
• Systems that connect model quality with measurable business outcomes such as production efficiency and conversion.

Earlier Research

• Computer vision, cross-modal retrieval, and visual-text matching.
• Medical AI and multimodal representation learning.
• Video understanding, retrieval, and structured content analysis.