Two words, anxious, focus, break out, sink down, runaway, keep stay, feel it, try it, crush it, make it, everybody move. —— imitating Two Words, Kanye West

Hello there! My name is Yijie Jin (金义杰). I am currently an undergraduate student at the Department of Computer Science and Technology, School of Computer Engineering and Science, Shanghai University (SHU), supervised by Prof. Junjie Peng.

My English name is Drew. In English-speaking contexts, I prefer to use Drew as my middle name, so my full name can be Yijie Drew Jin. Feel free to address me by any of these names—whether it’s Yijie, Drew, or Jin—I’m happy to respond to any of them!

I am passionate about advancing Machine Learning Systems (MLSys), with a strong focus on Efficient Large Language Models (LLMs). Additionally, I used to explore traditional NLP and multimodal learning, particularly in the domain of Multimodal Sentiment Analysis (MSA). My first paper, which integrates these research areas (both MLSys and MSA), has been submitted to ACL 2025 and is currently under review, with a meta score 4 in ARR Dec. Round.

Recently,

• I have been serving as a Research Intern with Prof. Zhijie Deng at Shanghai Jiaotong University (SJTU), diving into Edge LLM deployment with all sorts of LLM reasoning & inference optimization.

• Currently learning all kinds of stacks:
  • Lang: Triton, CPP (keeps on learning, diving in), Golang in backend development, Vue in frontend development.
  • MLSys/Efficient LLMs: parallel decoding, efficient reasoning, kv compression, etc. (hard! hard! hard!)
  • LLM Reasoning: all kinds of reasoning tasks, especially reasoning pruning.

• Planning to learn: Ray, TileLang, practice my leetcode ablitiy, etc.

• Projects planning to work on: minitorch.

🤔 Skills

• Lang: Skillful Python; Naive C/CPP/CUDA; Learning Golang, JS/TS

• Frameworks: Skillful PyTorch; Naive LibTorch, JAX + Haiku, Triton; Learning Gin + Gorm, Vue

• Tools: All toolchains in Visual Studio Code; vscode-clangd-CMake-toolchain, Git, Linux, MacOS (UNIX), Docker

🤓 Blogs

All my blogs are posted on Zhihu. Feel free to follow me: Drew.

I regularly update my blog with a variety of content, including experience sharing, technical coding tutorials, academic paper discussions, personal career insights, and casual life reflections—all written in a relaxed and conversational style.

🔥 News

• 2025.05:  🎉🎉 GsiT accepted by ACL 2025 (main).

• 2025.05:  🎉🎉 Our team (SHU Optimus) make it NO.7 Worldwide in the Final Round of ASC 2025 Student Supercomputer Challenge (ASC25).

• 2025.03:  🎉🎉 Our team (SHU Optimus) win the First Prize (NO.5 Worldwide) in the Preliminary Round of ASC 2025 Student Supercomputer Challenge (ASC25).

📝 Publications

ACL 2025 (Main)

Multimodal Transformers are Hierarchical Modal-wise Heterogeneous Graphs

Yijie Jin, Junjie Peng, Xuanchao Lin, Haochen Yuan, Lan Wang, Cangzhi Zheng

Project

• Efficiency and Effectiveness: 1/3 number of parameters without additional computational overhead compared with traditional MulTs.
• Modal-wise Block-Sparse Attention and All-Modal-In-One Fusion.

🎖 Honors and Awards

• 2025.03 First Prize of 2025 Asia Supercomputer Community Student Supercomputer Challenge (ASC25, NO.5 Worldwide Preliminary Round, NO.7 Worldwide Final Round).

• 2024.11 Second Prize of Shanghai Area in The 2024 China Undergraduate Mathematical Contest in Modeling (CUMCM 2024).

• 2024.11 Shanghai University Academic Excellence First-class Scholarship.

• 2024.08 First Prize of The 14th Asia and Pacific Mathematical Contest in Modeling (APMCM 2024).

• 2023.11 Shanghai University Academic Excellence Top-class Scholarship (Rank 15 in 1500).

📖 Educations

• 2022.09 - (now), Undergrad., Shanghai University.

• 2019.09 - 2022.07, Senior High., The Affiliated High School of Sichuan University (Chengdu NO.12 High School).

• 2016.09 - 2019.07, Junior High., Chengdu No.7 High School (Chengdu High School).

• 2010.09 - 2016.07, Primary School, Chengdu Paotongshu Primary School.

💬 Invited Talks

• Still working on it.

💻 Internships

• 2025.03 - (now), Research Intern, SJTU DengLab, Shanghai Jiaotong University, Shanghai, China.

• 2024.10 - 2024.12, Research Intern, NLP Group, Hongkong University of Science and Technology Guangzhou (HKUST-gz), Guangzhou, China.

• 2023.11 - (now), Research Assistant, LITS, Shanghai University, Shanghai, China.