Applied Researcher II
eBay · Shanghai, China
Applied research for search science, multimodal item understanding, generative listing assistance, and retrieval systems.
Focus
Language Models
Vision-Language Models
Multimodal Retrieval
Large-scale AI Systems
Search & E-commerce
Highlights
- Built production-grade multimodal and generative AI systems for e-commerce search, item understanding, and listing quality.
- Experienced across the full applied research loop: problem framing, data construction, model training, evaluation, retrieval, inference, and system integration.
- Published peer-reviewed research and contributed to multiple patent filings in search, attribute generation, and multimodal data processing.
Experience
Machine Learning Engineer
ZTE Corporation · Shanghai, China
Machine learning engineering for model development and applied AI systems.
Radio Frequency Engineer
HiSilicon · Shanghai, China
RF engineering after graduate research in electronic science and technology.
Education
Shanghai Jiao Tong University
M.S. in Electronic Science and Technology
National-level scholarship; strong academic standing
Harbin Institute of Technology
B.E. in Electromagnetic Field and Wireless Technology
Strong academic standing
Work Themes
Multimodal Understanding
Designed vision-language pipelines that extract structured product signals from images and text.
Improved attribute coverage and downstream search quality while balancing precision, latency, and operational reliability.
Generative Listing Assistance
Built customized generative models and preference-aligned training flows for long-tail content completion.
Expanded intelligent suggestion coverage and improved listing quality in production scenarios.
Multimodal Retrieval
Developed image-text representation learning and vector retrieval pipelines for similar-item discovery.
Supported large-scale recommendation and visual shopping use cases with production retrieval systems.
Semantic Recall
Trained semantic item retrieval models using business-oriented datasets and hard-negative strategies.
Strengthened recall quality for sparse and low-result search scenarios.
Research
- 6 SCI/EI-indexed publications in earlier research work
- Multiple patent contributions related to search and multimodal data processing
Skills
Core domains
Language Models · Vision-Language Models · Multimodal Learning · Vector Retrieval · Representation Learning · Foundation Model Training · Preference Alignment · Production AI Systems
