Woofun AI reports that YuXian Gu, a Tsinghua University computer science Ph.D. graduate and winner of the 2025 Outstanding Graduate Student Award, has officially joined DeepSeek. His name is now listed as an author on the DeepSeek V4 paper.

Gu’s research focuses on optimizing large model efficiency in pre-training, compression, and inference, with nearly 5,000 Google Scholar citations. His previous notable works include MiniLLM, adopted by Google, Alibaba, and NVIDIA, and the hybrid architecture model Jet-Nemotron. On an H100 GPU processing 256K ultra-long contexts, Jet-Nemotron achieves a throughput speedup of up to 53.6 times compared to traditional full-attention models.