Shangyan Zhou
Shangyan Zhou is an AI infrastructure builder whose work connects large language models with the systems behind training and inference. He graduated from Peking University in 2020 and spent three years at High-Flyer before joining DeepSeek. He coauthored the DeepSeek LLM, DeepSeek-V2, DeepSeek-V3 and DeepSeek-R1 papers, plus DeepEP, an expert-parallel communication library. His 2025 work also examines the hardware and infrastructure behind DeepSeek-V3 and R1. As of October 2026, he is an infrastructure developer at DeepSeek.
Timeline
-
2020
Peking University graduation
He graduates from Peking University in 2020.
-
January 2024
DeepSeek LLM
In January 2024, he coauthors DeepSeek LLM, a paper that studies scaling laws for open-source models at 7B and 67B sizes. The team trains its chat models with supervised fine-tuning and direct preference optimization.
-
May 2024
DeepSeek-V2 architecture
In May 2024, he coauthors DeepSeek-V2, a 236-billion-parameter mixture-of-experts model with 21 billion active parameters per token. Its architecture combines Multi-head Latent Attention with DeepSeekMoE for efficient inference and economical training.
-
December 2024
DeepSeek-V3 Technical Report
In December 2024, he coauthors the DeepSeek-V3 technical report, which presents a 671-billion-parameter mixture-of-experts model with 37 billion active parameters per token. The model trains on 14.8 trillion tokens.
-
January 2025
DeepSeek-R1 reasoning model
In January 2025, he coauthors DeepSeek-R1, which uses reinforcement learning to develop reasoning abilities such as self-reflection, verification and dynamic strategy adaptation. The method uses no human-labeled reasoning trajectories.
-
January 2025
Infrastructure work at DeepSeek
By January 2025, he works as a DeepSeek infrastructure developer. Before joining DeepSeek, he spends three years at High-Flyer, Liang Wenfeng's hedge fund.
-
2025
DeepEP and infrastructure analysis
In 2025, he coauthors DeepEP, an efficient expert-parallel communication library. He also coauthors a study of DeepSeek-V3 and R1, analyzing their architecture, AI infrastructure and hardware choices.
-
As of October 2026
Infrastructure developer at DeepSeek
As of October 2026, he is an infrastructure developer at DeepSeek.
Career
- NowDeepSeekInfrastructure developer
Education
Sources and credits
Photo: We have no published photo of Shangyan Zhou yet. Send us a free-licensed photo.
- Deepseek's Liang Wenfeng hires graduates over "AI Wizards"
- DeepSeek LLM: Scaling Open-Source Language Models with Longtermism
- DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
- DeepSeek-V3 Technical Report
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectures
- DeepEP: an efficient expert-parallel communication library
- DeepSeek AI | The Org
