Zhanheng Nie
Portrait of Zhanheng Nie

Zhanheng Nie

M.S. Student, Beihang University

I study how multimodal models learn useful representations and keep adapting as data, tasks, and contexts change.

"Zero to start, and one is the best motivation."

I am actively seeking PhD opportunities for Fall 2027 in Hong Kong.

Research Topics

Multimodal Representation Learning

Building generative MLLM-based models for product understanding at industrial scale

Continual & Adaptive Learning

Enabling models to learn from unbounded data streams without catastrophic forgetting

Efficient Mixture Architectures

Designing hierarchical MoE systems that unify recognition, generation, and representation

Education

Beihang University

M.S., Computer Science

Sep 2024 — Jun 2027

Beihang University

B.Eng., Software Engineering

Sep 2019 — Jun 2024

Internship

Alibaba Group

Research Intern, Search Ads

Apr 2025 — Present

View Publications

News

Jul. 2026
📄 Research

1 paper accepted by ACM MM 2026 (co-first author)

Jun. 2026
📄 Research

1 paper published at CVPR 2026 (co-first author, independently led)

Feb. 2026
📄 Research

1 paper published at WSDM 2026 Oral (co-first author)

Nov. 2025
📄 Research

MOON Technical Report (31 pages) released on arXiv, documenting +20% CTR deployment

Apr. 2025
💼 Work

Joined Alibaba Group (Taobao & Tmall) as Research Intern in Search Advertising

Sep. 2024
🏆 Award

Started M.S. at Beihang University, School of CS&E (Top 20%)

Jun. 2024
🏆 Award

Outstanding Graduate of Beihang University (Top 10%)

Publications

* denotes equal contribution. Google Scholar

MOON2.0: Dynamic Modality-balanced Multimodal Representation Learning for E-commerce Product Understanding

MOON2.0: Dynamic Modality-balanced Multimodal Representation Learning for E-commerce Product Understanding

TL;DR: We propose Modality-Driven MoE and Dual-Level Alignment to solve modality imbalance in multimodal product representation, achieving SOTA on multiple benchmarks.

Zhanheng Nie*, Chenghan Fu*, Daoze Zhang*, Junxian Wu*, et al.

Published Poster CVPR 2026 Co-first author (1st), independently led all work
Paper Dataset
MOON: Generative MLLM-based Multimodal Representation Learning for E-commerce Product Understanding

MOON: Generative MLLM-based Multimodal Representation Learning for E-commerce Product Understanding

TL;DR: We propose the first generative MLLM-based product representation model with Guided MoE and specialized negative sampling.

Daoze Zhang*, Chenghan Fu*, Zhanheng Nie*, Jianyu Liu*, et al.

Published Oral WSDM 2026 Co-first author
Paper Dataset
MOON3.0: Reasoning-aware Multimodal Representation Learning for E-commerce Product Understanding

MOON3.0: Reasoning-aware Multimodal Representation Learning for E-commerce Product Understanding

TL;DR: We propose the first reasoning-aware MLLM for product representation, combining contrastive learning with GRPO reinforcement and a FIRE module for fine-grained details.

Junxian Wu*, Chenghan Fu*, Zhanheng Nie*, Daoze Zhang*, et al.

Accepted Poster ACM MM 2026 Co-first author
Paper
MOON Embedding: Multimodal Representation Learning for E-commerce Search Advertising

MOON Embedding: Multimodal Representation Learning for E-commerce Search Advertising

TL;DR: A comprehensive 31-page report documenting MOON deployment across Taobao search advertising, achieving cumulative +20% CTR over three years.

Chenghan Fu*, Daoze Zhang*, Yukang Lin*, Zhanheng Nie*, et al.

Technical Report arXiv:2511.11305 Co-first author
Paper

Experience

Apr 2025 — Present

Research Intern / LLM Algorithm Engineer

Alibaba Group, Taobao & Tmall Search Advertising

Core contributor to the MOON series of multimodal product representation models. Deployed across Taobao search advertising pipeline (recall, relevance, ranking), achieving cumulative +20% CTR — the largest single-project lift in search ads over three years.

Sep 2024 — Present

Graduate Researcher

Beihang University, School of CS&E

Research on continual learning and mixture-of-experts architectures under Prof. Jia Li. Published at CVPR, WSDM, and ACM MM. Investigating hierarchical MoE approaches for online continual learning and continual personalized generation.

Sep 2019 — Jun 2024

B.Eng. in Software Engineering

Beihang University, School of Software

GPA 3.82/4.0 (Top 6%). National Encouragement Scholarship recipient. First Prize in National Mathematical Modeling Contest (Beijing).

Selected Awards

2026 CVPR 2026 Publication (Poster)
2026 ACM MM 2026 Publication (Poster)
2026 WSDM 2026 Publication (Oral)
2024 Outstanding Graduate, Beihang University
2022-2023 National Encouragement Scholarship (Top 10%)
2020, 2022 First Prize, National Mathematical Modeling Contest (Beijing)
2022 First Prize, FLTRP English Public Speaking Contest (Provincial)
2020 Honorable Mention, MCM/ICM (USA)

About

I am a third-year M.S. student at the School of Computer Science and Engineering, Beihang University, advised by Prof. Jia Li. My research focuses on multimodal representation learning and continual learning, with Mixture-of-Experts as a unifying architectural lens. I am currently a research intern at Alibaba Group (Taobao & Tmall), where I contribute to the MOON series of multimodal product representation models deployed in search advertising.

I am actively seeking PhD opportunities for Fall 2027. If you are interested in collaboration or have openings in related research areas, please feel free to reach out.

Contact & Links