PeopleofSuperintelligence

Researcher

Chengqi Deng

Also known as KinglittleQ

Chengqi Deng is an AI researcher whose work spans singing voice conversion, mixture-of-experts language models, and systems for model training and inference. He co-authored PitchNet at Zhejiang University and later co-authored DeepSeekMoE, DeepSeek-V3, and DeepSeek-R1. His work also includes DeepEP, an expert-parallel communication library, a hardware analysis of DeepSeek-V3, DeepSeek-V4, and the DSpark decoding method. As of October 2026, he works on DeepSeek-V4.1-Flash and researches similarity search and voice conversion.

Works
20
Citations
3,068
h-index
13
Badges

Publication numbers from OpenAlex, as of 19 October 2026.

Timeline

  1. 2019–2020

    PitchNet for singing voices

    He submits PitchNet in 2019, an unsupervised singing voice conversion method that adds pitch regression for better pitch control. The 2020 ICASSP program lists him with Zhejiang University for the paper.

  2. 2024

    Expert specialization in MoE

    He co-authors DeepSeekMoE, a mixture-of-experts language model built around fine-grained expert segmentation and shared expert isolation.

  3. 2024–2025

    DeepSeek-V3 and hardware design

    His 2024 DeepSeek-V3 report presents a 671-billion-parameter mixture-of-experts model with 37 billion parameters active per token. At ISCA in 2025, he co-authors an analysis of its hardware design, including FP8 training and multi-plane networking.

  4. January 2025

    Reasoning with DeepSeek-R1

    In January 2025, he co-authors DeepSeek-R1, a paper that shows large language models can develop reasoning abilities through reinforcement learning.

  5. 2025

    Expert-parallel communication

    He co-authors DeepEP, a library for expert-parallel communication in machine learning training and inference, in 2025.

  6. April 2026

    DeepSeek-V4 models

    In April 2026, he co-authors the DeepSeek-V4 report, which presents V4-Pro and V4-Flash with context lengths of up to one million tokens.

  7. July 2026

    Confidence-scheduled decoding

    In July 2026, he co-authors DSpark, a speculative decoding method that uses confidence-scheduled verification to improve generation efficiency.

  8. As of October 2026

    DeepSeek research and engineering

    He is part of the DeepSeek-V4.1-Flash research and engineering team and works on a multimodal model that supports one-million-token contexts. He is a member of DeepSeek-AI and ZJULearning and researches similarity search and voice conversion.

Career

  1. Since 2019Zhejiang UniversityMember
  2. NowDeepSeek AIMember
  3. 2024–2026DeepSeek

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning2025 · Nature · 1,112 citationsWith Bingxuan Wang, Bochao Wu, Damai Dai, Daya Guo and Zhibin Gou
  2. Can Open Large Language Models Catch Vulnerabilities?2025 · 538 citationsWith Bingxuan Wang, Bochao Wu, Damai Dai and Zhibin Gou
  3. The Faiss Library2025 · IEEE Transactions on Big Data · 460 citationsWith Hervé Jégou and Matthijs Douze
  4. DeepSeek-V3 Technical Report2024 · 292 citationsWith Bingxuan Wang, Bochao Wu, Damai Dai, Daya Guo and Dejian Yang
  5. DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models2024 · Annual Meeting of the Association for Computational Linguistics (ACL) · 282 citationsWith Damai Dai and Zhenda Xie

From OpenAlex, as of 19 October 2026.

  1. The median h-index is 53 at universities and 19 at frontier AI labsIncludes Chengqi Deng. The median h-index is 53 at universities and 19 at frontier AI labs. We compare publishing careers, citations and moves to estimate how universities and frontier labs contribute to AI research.
  2. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Zhejiang University. Our estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.
  3. The company share of AI researchers’ top papers rose from 2% to 32%Includes Zhejiang University. We estimate where leading AI researchers worked when their most-cited papers appeared. The company share rose from 2% in 1975–1979 to 32% in 2020–2024.
  4. Google has the highest combined h-index among AI organizationsIncludes Zhejiang University. Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Chengqi Deng yet. Send us a free-licensed photo.

  1. PitchNet: Unsupervised Singing Voice Conversion with Pitch Adversarial Network arxiv.org. Checked 19 October 2026.
  2. IEEE ICASSP 2020 || Barcelona, Spain || 4-8 May 2020 cmsworkshops.com. Checked 19 October 2026.
  3. DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models arxiv.org. Checked 19 October 2026.
  4. DeepSeek-V3 Technical Report arxiv.org. Checked 19 October 2026.
  5. Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectures arxiv.org. Checked 19 October 2026.
  6. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning arxiv.org. Checked 19 October 2026.
  7. GitHub - deepseek-ai/DeepEP: DeepEP: an efficient expert-parallel communication library github.com. Checked 19 October 2026.
  8. DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence arxiv.org. Checked 19 October 2026.
  9. DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation arxiv.org. Checked 19 October 2026.
  10. DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression arxiv.org. Checked 19 October 2026.
  11. KinglittleQ (Chengqi Deng) · GitHub github.com. Checked 19 October 2026.