🎓 MS in Computer Science at UC San Diego (2025-present)

• BS in Computer Science at ShanghaiTech University (2021-2025)
• Exchange student at UC Berkeley Extension (GLOBE Program) with 4.0/4.0 GPA

🔬 Research Experience

• Shanghai Alibaba Ant Group NLP Lab (Jul. 2025-Present): Researching Multi-Head FFN as a powerful and efficient alternative to FFNs in Transformers.
• ShanghaiTech Kewei Tu's Lab (Feb. 2025-Present): Using sparse autoencoders to probe entity-specific knowledge within LLMs.
• Shanghai Qizhi Institute (Jun. 2024-Jan. 2025): Developed an analytical solution for real-time inverse-refraction problems.

📝 Publications

• Flash Multi-Head Feed-Forward Networks (Under Review)
  • Minshen Zhang*, Xiang Hu*, Jianguo Li, Wei Wu, Kewei Tu
  • FlashMHF consistently improves perplexity and downstream task accuracy over SwiGLU FFNs, while reducing peak memory usage by 3-5x and accelerating inference by up to 1.08x.

🔭 I'm currently working on

• Flash Multi-Head FFN and other novel LLM structures
• Interesting problems in Machine Learning systems.
• Retrieval based Long-context end-to-end language modeling systems.

🌱 Skills & Technologies

• MLsys: PyTorch, Triton, ThunderKittens
• CUDA C++: Implemented custom Flash Algorithm on Hopper with warp specialization and WMMA.
• Libraries: Hugging Face Transformers (Authored a merged PR)
• Languages: Python, C/C++, C

🏆 Honors

• Outstanding Graduate of ShanghaiTech University (2024-2025)
• Outstanding Student of ShanghaiTech University (2021-2022, 2023-2024)

👨‍🏫 Teaching: Teaching Assistant for Computer Programming (CS100) at ShanghaiTech University (Spring 2024)