Andy's CV
The PDF format is available here.
Education
MMath, Computer Science - University of Waterloo, Ontario, Canada
Sept 2024 – Now
MComp, Artificial Intelligence - National University of Singapore, Singapore
Sept 2022 – May 2024
- Advised by Presidential Young Professor Yang You
BEng, Data Science - Lanzhou University, Lanzhou, China
Sept 2018 – May 2022
- GPA: 92.8/100 (Ranking: 1/192)
Research Projects
OpenMoE: An Early Effort on Open Mixture-of-Experts Language Models
- OpenMoE is the the first open-source, decoder-only MoE language model, providing insights about the routing mechanism to the open-source community (Huggingface).
- Prepared the training dataset, tokenizer and conducted the model evaluation. Worked on the PyTorch implementation with teammates.
Matrix: Infinite-Horizon World Generation with Real-Time Interaction
- Built a data collection pipeline for Cyberpunk 2077, recording per-frame video data and corresponding control signals to support model training.
- Served as a core contributor to a multi-GPU inference framework, implementing Ray-based workers (DiT, VAE, post-processing) and building an interactive frontend/backend system.
- Delivered the first real-time game generation demo with user-controllable inference (demo).
AdaVocab: Boosting SLM Inference with Sparse Vocabulary Activation
- Identified the growing vocabulary size as a major bottleneck for the Small Language Model (SLM) inference efficiency.
- Proposed and implemented a sparsely active vocabulary method; prepared training data, modeled Trainer, and completed evaluation with teammates.
- Achieved over 20% computation reduction and 10% inference speedup for SLMs in CPU settings.
Work Experience
HPC-AI Tech, Artificial Intelligence Engineer Intern
Beijing, China | July 2023 – Nov 2023
- Extended LLaMA’s vocabulary for Chinese and contributed to data preparation in the Colossal-LLaMA-2 project, selected as an official base model in the 2023 NeurIPS LLM Efficiency Challenge.
- Investigated common context length extrapolation methods (e.g., PI, NTK, LongLoRA), and implemented corresponding training and evaluation pipelines to extrapolate Colossal-LLaMA-2 with multi-GPU training.
- Worked on the ColossalQA project, a RAG framework based on Langchain.
Publications
[ICML 2024] OpenMoE: An Early Effort on Open Mixture-of-Experts Language Models
Fuzhao Xue, Zian Zheng, Yao Fu, Jinjie Ni, Zangwei Zheng, Wangchunshu Zhou, Yang You
Jan 2024
[ICLR 2025] MixEval-X: Any-to-Any Evaluations from Real-World Data Mixtures
Jinjie Ni, Yifan Song, Deepanway Ghosal, Bo Li, David Junhao Zhang, Xiang Yue, Fuzhao Xue, Zian Zheng, Kaichen Zhang, Mahir Shah, Kabir Jain, Yang You, Michael Shieh
Oct 2024
[Arxiv] The Matrix: Infinite-Horizon World Generation with Real-Time Moving Control
Ruili Feng, Han Zhang, Zhantao Yang, Jie Xiao, Zhilei Shu, Zhiheng Liu, Andy Zheng, Yukun Huang, Yu Liu, Hongyang Zhang
Dec 2024
Teaching & Service
- Teaching Assistant: CS135, CS479/679 at UWaterloo
- Reviewers: ICLR 2025, AISTATS 2025
Honors & Awards
- China National Scholarship (Top 0.1% across nation), Dec 2019 & Dec 2021
- Merit Student in Colleges and Universities in Gansu Province (Top 1% across province), May 2021
- Dr. Derick Wood Graduate Scholarship, Dec 2024