Zhihong Shao (邵智宏)

Core Researcher (R1, V2, DeepSeek-Math, DeepSeek-Prover, Math-Shepherd) — DeepSeek

Website Google ScholarOpenReview

Research Scientist at DeepSeek focused on LLM reasoning and self-improving systems. Core contributor to DeepSeek-R1 (published in Nature 2025), lead/first author of DeepSeekMath (pushing mathematical reasoning limits in open models), and co-author of DeepSeek-Prover (theorem proving), Math-Shepherd (step-by-step verification without human annotations, ACL 2024), and ToRA (ICLR 2024, tool-integrated reasoning, interned at MSRA).

PhD from Tsinghua University (2019–2024, advisor Minlie Huang). BS from Beihang University (multiple national scholarships, math competition awards). MIT Technology Review "35 Under 35" (2024). NeurIPS Area Chair.

Affiliations

  • ●
    DeepSeek — Core Researcher (R1, V2, DeepSeek-Math, DeepSeek-Prover, Math-Shepherd)
    Formerly: Tsinghua University (PhD 2019-2024, advisor Minlie Huang); BS Beihang University; MSRA intern (ToRA)