MiniMax AI Research Team Contributors | Generated by AI
Key Contributors from MiniMax AI Research Team
The listed individuals are primarily members of MiniMax AI’s research and engineering team, as evidenced by their co-authorship on key technical papers like the MiniMax-M1 report (arXiv:2506.13585) and related works on long-context reasoning, multimodal models, and agentic systems. MiniMax, founded in December 2021, emphasizes efficient scaling of foundation models toward AGI, with a core team drawn from veterans of companies like SenseTime. Public details on individual roles are limited, as the company maintains a low-profile approach focused on technical output rather than personal bios. Below is a summary based on available affiliations, publications, and professional backgrounds:
- Haoyu Feng: AI Researcher at MiniMax. Contributed to core architecture and attention mechanisms in MiniMax-M1.
- Huaiguang Cai: AI Researcher at MiniMax. Focused on reinforcement learning and test-time compute optimizations in model training pipelines.
- Haichao Zhu: AI Engineer at MiniMax. Involved in hybrid MoE (Mixture-of-Experts) implementations for efficient scaling.
- Jian Sun: AI Researcher at MiniMax. Specializes in vision-language integration, with prior work on datasets like Objects365.
- Jiaqi Zhuang: AI Researcher at MiniMax. Contributed to data synthesis and verifiable reasoning frameworks like SynLogic.
- Jiaren Cai: AI Engineer at MiniMax. Worked on long-context handling and Lightning Attention for 1M+ token windows.
- Jiayuan Song: AI Researcher at MiniMax. Expertise in multimodal entity tracking and evaluation benchmarks (e.g., MET-Bench).
- Jin Zhu: AI Researcher at MiniMax. Focused on super-resolution and inference-time scaling techniques.
- Jingyang Li: AI Engineer at MiniMax. Contributed to positional encoding and rotary embeddings in MiniMax-01 series.
- Jinhao Tian: AI Researcher at MiniMax. Involved in RLHF (Reinforcement Learning from Human Feedback) for agentic workflows.
- Jinli Liu: AI Engineer at MiniMax. Specialized in efficient training on limited hardware (e.g., H800 GPUs).
- Junhao Xu: AI Researcher at MiniMax. Worked on multi-agent collaboration (MCP) architectures.
- Junjie Yan: Founder and CEO of MiniMax. Former Vice President and Head of R&D at SenseTime; oversees technical direction and AGI roadmap. Key visionary behind models like MiniMax-M1 and M2.
- Junteng Liu: PhD Candidate in NLP at Hong Kong University of Science and Technology (HKUST); former Research Intern at MiniMax (June–September 2024). Co-authored SynLogic and MiniMax-M1; focuses on logical reasoning data synthesis.
- Junxian He: Assistant Professor at HKUST (Machine Learning & NLP focus, 8,500+ citations). Collaborator with MiniMax on reasoning models; affiliated contributor to MiniMax-M1 and SynLogic papers.
- Kaiyi Feng: AI Researcher at MiniMax. Contributed to test-time adaptation and extrapolation techniques (e.g., e3 framework).
- Ke Yang: AI Engineer at MiniMax. Specialized in scalable RL training and reward modeling for diverse tasks.
These individuals form part of MiniMax’s ~200-person team, heavily skewed toward R&D (per company profiles). The broader founding team includes co-founders Yang Bin (Technical Partner, ex-Uber AI and autonomous driving expert) and Zhou Yucong, both SenseTime alumni. For deeper insights, MiniMax’s open-source repos (e.g., GitHub/MiniMax-AI) highlight collaborative contributions without granular role breakdowns.
References
- MiniMax-M1 Technical Report
- MiniMax Company Wikipedia
- Yan Junjie Profile (Crunchbase)
- Junxian He Google Scholar
- Junteng Liu Personal Site
- MiniMax About Page
- Zhihu: MiniMax Team Overview