Education
Fudan University
Shanghai · Sep 2023–present
Undergraduate in Artificial Intelligence
Integrated undergraduate–graduate track · Computer Science class
01 / profile
A short introduction.
Grade
大三 / Year 3
As listed on my current résumé
Focus
Multimodal AI
Video understanding · data engineering · evaluation tooling
Toolbox
02 / experience
Research, from data to demo.
Research intern · Streaming Video Understanding Team
Video-understanding large models & data engineering
- Designed a data pipeline around PB-scale capacity; the delivered filtering workflow improved speed by 100% over the previous process.
- Built six Vue 3 + Flask internal applications for data checks, training submission, inference QA, resource management, and interactive demos.
- Supported approximately 2M OCR-grounding captions, 1M identity-cognition records, and 60M fine-grained text labels.
Research intern · Large-model NLP data construction & cleaning
Building usable datasets from the ground up
- Reproduced an MBTI personality dialogue dataset and expanded it with structured role-card collections.
- Implemented sensitive-word data cleaning and quality checks for a high-quality emotional-dialogue database.
03 / projects
Selected repositories.
01Vision_LLMLLaVA-style vision-language reproduction
02NieraSpeech assistant experiment · ASR / TTS
03MM_Dataset_CltMultimodal dataset collection workflow
Project links point to public repositories; descriptions are intentionally brief and résumé-like.
04 / paper
MOSS-Video-Preview.
Toward Real-Time Video Understanding via Cross-Attention
Co-author · 2026 technical report · Fudan University (OpenMOSS project)
A cross-attention video-language foundation model designed for streaming and low-latency interaction, keeping visual perception and text generation concurrent.
05 / contact
Let’s stay in touch.
For research conversations, collaborations, or a quick hello: