Kaibin Tian

Algorithm Engineer
TikTok
tikibi@foxmail.com

About Me

I am an algorithm engineer at TikTok. I received my M.S. in Computer Science from Renmin University of China (RUC) in 2023, advised by Prof. Xirong Li at AIMC Lab. Before that, I received my B.E. in Software Engineering from University of Electronic Science and Technology of China (UESTC) in 2020. Prior to joining TikTok, I worked as an algorithm engineer at Kuaishou Technology (2023–2025). I have also been a research intern at SenseTime and Kuaishou Technology during my graduate studies.

Research Interests

Video Understanding
Cross-Modal Retrieval
Reinforcement Learning
Multimodal Large Language Model

News

[2025] We open-sourced SeekWorld, a tool designed for geolocation reasoning tasks. ⭐ 60+
[2025] Our paper “IFSR” about preference alignment of LLM is accepted to EMNLP Findings 2025.
[2024] Our paper “TeachCLIP” about text-to-video retrieval is accepted to CVPR 2024.
[2024] Our paper “EERCF” about text-to-video retrieval is accepted to AAAI 2024.
[2023] Our paper “ChinaOpen” (a new video dataset targeted at open-world multimodal learning) is accepted to ACM MM 2023.

Publications

ICML

BubbleSpec: Turning Long-Tail Bubbles into Speculative Rollout Drafts for Synchronous Reinforcement Learning

Yuhang Xu*, Kaibin Tian*, Yang Tian, Zhice Yang, Yifeng Yu, Yan Li, Shengzhong Liu, Fan Wu, Guihai Chen

International Conference on Machine Learning (ICML), 2026.

PDF

EMNLP

Improving Preference Alignment of LLM with Inference-Free Self-Refinement

Fukun Ma, Kaibin Tian, Jieting Xue, Xiaoyi Wang, Ye Ma, Quan Chen, Peng Jiang, Lijie Wen

Findings of the Association for Computational Linguistics (EMNLP Findings), 2025.

PDF

CVPR

TeachCLIP: Multi-Grained Teaching for Efficient Text-to-Video Retrieval

Kaibin Tian, Ruixiang Zhao, Hu Hu, Runquan Xie, Fengzong Lian, Zhanhui Kang, Xirong Li

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024.

PDF Code

AAAI

Towards Efficient and Effective Text-to-Video Retrieval with Coarse-to-Fine Visual Representation Learning

Kaibin Tian, Yanhua Cheng, Yi Liu, Xinglin Hou, Quan Chen, Han Li

AAAI Conference on Artificial Intelligence (AAAI), 2024.

PDF Code

ACMMM

ChinaOpen: A Dataset for Open-world Multimodal Learning

Aozhu Chen, Ziyuan Wang, Chengbo Dong, Kaibin Tian, Ruixiang Zhao, Xun Liang, Zhanhui Kang, Xirong Li

ACM International Conference on Multimedia (ACM MM), 2023.

PDF Code

arXiv

Co-Teaching for Unsupervised Domain Adaptation and Expansion

Kaibin Tian, Qijie Wei, Xirong Li

arXiv preprint, 2022.

PDF Code

TOMM

Unsupervised Domain Expansion for Visual Categorization

Jie Wang*, Kaibin Tian*, Dayong Ding, Gang Yang, Xirong Li (*Equal contribution)

ACM Transactions on Multimedia Computing, Communications and Applications (TOMM), 2021.

PDF Code

Open Source Projects

SeekWorld: Geolocation is a Natural RL Task for o3-like Visual Clue-Tracking Reasoning.
Code Model

Competitions

2022 NIST TREC Video Retrieval Evaluation: Ad-hoc Video Search Task — Rank 2.
Note
2022 CVPR Workshop Challenge: Semi-Supervised Action Recognition In The Dark — Rank 2.
Poster Code
2022 CVPR Workshop Challenge: AICITY Naturalistic Driving Action Recognition — Rank 9.
Note Code
Alibaba Tianchi: Detection of Tampered Images — Rank 16/1666 (Top 1%).
Code

Services

Conference Reviewers

AAAI, NeurIPS, CVPR, ECCV

Journal Reviewers

ACM TOMM

Powered by Jekyll and Minimal Light theme.