PeopleofSuperintelligence

Researcher

Wei Du

Wei Du is a Research Scientist at NVIDIA. In November 2025, he focused on mathematical reasoning on the NeMo LLM team and maintained NeMo-Skills. He earned a computer science Ph.D. at the University of Arkansas in 2022, with dissertation research on robust and fair machine learning; his first job was as a data scientist at Qualtrics, and he later worked at Amazon. In 2024, he coauthored an NVIDIA post on LoRA adapters, and his 2025 work covered OpenMathReasoning, few-example reasoning and Nemotron-Math. In 2026, he coauthored research on generative selection and an open Nemotron proof pipeline that scored 30 of 42 points at the IMO, reaching the gold-medal threshold.

Now at
NVIDIA
Works
2
Citations
0
h-index
0
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. January 2019–May 2022

    A Ph.D. at Arkansas

    From January 2019 to May 2022, he pursues a Ph.D. in computer science at the University of Arkansas's Social Awareness and Intelligent Learning Lab. His dissertation focuses on robust and fair machine learning under distribution shift.

  2. June 2024

    LoRA adapters with NIM

    In June 2024, he coauthors an NVIDIA Technical Blog post on deploying multiple LoRA adapters with NVIDIA NIM. The post explains how clients select an adapter by its model name for each inference request.

  3. April 2025

    OpenMathReasoning at AIMO-2

    In April 2025, he coauthors the AIMO-2 winning-solution report and its OpenMathReasoning dataset, which contains 540,000 unique math problems and 3.2 million long-reasoning solutions. The team combines code execution with generative selection to choose promising answers.

  4. July 2025

    Reasoning from twenty examples

    In July 2025, he coauthors a study that lightly fine-tunes Qwen2.5-32B on 20 long chain-of-thought examples from QwQ-32B-Preview. The resulting model outperforms Qwen2.5-Math-72B-Instruct.

  5. December 2025

    Building Nemotron-Math

    In December 2025, he coauthors Nemotron-Math, a dataset with 7.5 million solution traces across high, medium and low reasoning modes, with and without Python tool-integrated reasoning. A sequential bucketed strategy speeds fine-tuning at 128K context by two to three times.

  6. February 2026

    Generative selection for Best-of-N

    In February 2026, he coauthors Learning Generative Selection for Best-of-N, which trains small reasoning models through targeted reinforcement learning to select among candidate solutions.

  7. September 2026

    An IMO proof system

    In September 2026, he coauthors an open Nemotron pipeline for olympiad proofs. The system generates, verifies and refines natural-language proofs and scores 30 of 42 points at IMO 2026, reaching the gold-medal threshold.

  8. As of October 2026

    Research Scientist at NVIDIA

    As of October 2026, he is a Research Scientist at NVIDIA. In November 2025, he works on mathematical reasoning on the NeMo LLM team through large-scale data training and reinforcement learning, and maintains NVIDIA's open-source NeMo-Skills library for LLM training and deployment. Before NVIDIA, he works at Qualtrics and Amazon on scalable machine-learning and NLP systems; his first job is as a data scientist at Qualtrics.

Career

  1. Since 2025NVIDIAResearch Scientist
  2. QualtricsData Scientist

Education

  1. 2022University of Arkansas at FayettevillePh.D. in computer science

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Wei Du. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. Google has the highest combined h-index among AI organizationsIncludes NVIDIA. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes NVIDIA. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes NVIDIA. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

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

Sources and credits

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

  1. People | Social Awareness and Intelligent Learning Lab sail.uark.edu. Checked 22 October 2026.
  2. Dr. Wei Du visited SAIL sail.uark.edu. Checked 22 October 2026.
  3. Seamlessly Deploying a Swarm of LoRA Adapters with NVIDIA NIM developer.nvidia.com. Checked 22 October 2026.
  4. AIMO-2 Winning Solution: Building State-of-the-Art Mathematical Reasoning Models with OpenMathReasoning dataset arxiv.org. Checked 22 October 2026.
  5. The Challenge of Teaching Reasoning to LLMs Without RL or Distillation arxiv.org. Checked 22 October 2026.
  6. Nemotron-Math: Efficient Long-Context Distillation of Mathematical Reasoning from Multi-Mode Supervision arxiv.org. Checked 22 October 2026.
  7. Learning Generative Selection for Best-of-N arxiv.org. Checked 22 October 2026.
  8. An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics arxiv.org. Checked 22 October 2026.
  9. wedu-nvidia (Wei Du) · GitHub github.com. Checked 22 October 2026.