CV
Short CV. For the full and most up-to-date version, please download the PDF above.
Contact Information
| Name | Binjie Zhang |
| Professional Title | Ph.D. Candidate, Computer Science |
| binjie97@u.nus.edu |
Professional Summary
Ph.D. candidate at Show Lab, National University of Singapore. Research interests span LLM agents (harness, memory, tool use, self-evolution), vision-language and video understanding, world models, and lifelong / compatible representation learning.
Experience
-
2026 - 2026 Singapore
AI Research Intern
TikTok (ByteDance)
Sole owner of several LLM-agent research and production projects, end-to-end from design to deployment.
- Pivoted from a vision / world-model background to a large-scale LLM-agent stack within two weeks.
- Designed and implemented a declarative, skill-based agent harness with LLM semantic routing and progressive disclosure.
- Built an Agent-Native self-evolving training loop (Diagnose, Hypothesize, Experiment, Evaluate, Reflect) on top of internal training / data / monitoring / deployment platforms.
- Researched training-free optimization of agent memory (retrieval, write, eviction) for long-horizon tool-use agents.
- End-to-end ownership - design, implementation, cross-team integration, review, and production rollout.
-
2023 - Present Singapore
Ph.D. Researcher
Show Lab, National University of Singapore
Research on LLM agents, world / predictive models, and lifelong / compatible representation learning.
- Reflection-augmented policy optimization for tool-using agents (ReGRPO).
- Lifelong imitation learning for robotic manipulation (RCFC).
- Egocentric future prediction conditioned on hand trajectories.
-
2020 - 2023 Shenzhen, China
AI Research Intern
Tencent ARC Lab - Compatible Representation Learning
Backward-compatible representation learning and model upgrades for large-scale image and video retrieval systems.
- Four first-author papers (ICLR’22, IJCAI’22 long oral, AAAI’23, arXiv’23) defining the team’s backward-compatible retrieval line.
- Multi-GPU contrastive / CLIP-style training pipelines with regression-alleviating methods enabling hot-refresh upgrades without re-encoding billion-scale galleries.
- Winner of the Tencent Technology Breakthrough Award (2022).
-
2019 - 2020 Shenzhen, China
AI Research Intern
Tencent ARC Lab - Cross-Modality Video Understanding
Multimodal models for video-text retrieval and temporal grounding on large-scale video-language datasets; built the PyTorch data / training / evaluation codebase for the group’s multi-modal research.
Education
-
2023 - 2027 Singapore
Ph.D.
National University of Singapore (NUS)
Computer Science
- Advisor - Asst. Prof. Mike Zheng Shou (Show Lab).
- Research on LLM agents, vision-language, world models, and compatible representations.
-
2019 - 2022 Beijing, China
M.Eng.
Tsinghua University (THU)
Computer Science and Technology
- Advisor - Prof. Chun Yuan.
- Focus on compatible representation learning and cross-modal video understanding.
- Excellent Master’s Graduate in Beijing; Outstanding Master’s Thesis.
-
2015 - 2019 Shanghai, China
B.Eng.
East China University of Science and Technology (ECUST)
Information Engineering
- Cumulative rank 1 / 92.
Academic Interests
LLM Agents: agent harness, skill systems, long-horizon memory, tool use, multi-modal CoT, reflection-augmented RL, self-evolving training loops
Vision-Language and Video: multi-modal foundation models, video-text retrieval, temporal grounding, egocentric video understanding
World and Predictive Models: future prediction, hand-trajectory conditioning, latent diffusion, embodied planning
Lifelong and Compatible Representation Learning: backward-compatible retrieval, hot-refresh model upgrades, prototype replay, coarse-to-fine compatibility
Awards
-
2022 Tencent Technology Breakthrough Award
Tencent
For Hot-Refresh Model Upgrades.
-
2022 SZCCF Science and Technology Award
Shenzhen CCF
For efficient compatible model upgrades.
-
2022 Excellent Master Degree Graduate in Beijing & Outstanding Master's Thesis
Beijing Municipal Education Commission / Tsinghua University
-
2018 Annual College Personage Award
ECUST
Highest student honor at ECUST.
-
2017 National Scholarship for Undergraduates (twice)
Ministry of Education of China
2016 and 2017.
Skills
Research Areas: LLM agents (harness / memory / tool use / RL), vision-language models, world and predictive models, video understanding, lifelong and compatible representation learning
Languages and Tools: Python (primary), C/C++, SQL, Bash, LaTeX
ML and Agent Stack: PyTorch, HuggingFace, DeepSpeed, vLLM, LangGraph / LangChain, MCP, CLIP, Latent Diffusion
Infra: distributed multi-GPU training, sandboxed code execution, large-scale training platforms
Academic Service
- Reviewer: ECCV, ICCV, CVPR, ICLR, AAAI, IJCAI.