PeopleofSuperintelligence

Lab leader

Maarten Van Segbroeck

Also known as maartenvs

As of October 2026, Maarten Van Segbroeck is director of research at NVIDIA, where he leads initiatives in synthetic data, privacy, and trustworthy AI. In 2006 he co-authored an Interspeech paper on missing-data speech recognition, and he earned a doctorate in electrical engineering from KU Leuven. Before joining NVIDIA, he served as VP of applied science at Gretel. In December 2025 he co-authored the Nemotron 3 Nano paper, and in September 2026 he co-authored NeMo Data Designer, an open-source framework for multi-modal synthetic data generation.

Now at
NVIDIA
Works
51
Citations
977
h-index
18

Publication numbers from OpenAlex, as of 15 October 2026.

Timeline

  1. 2006

    Missing-data speech recognition

    He co-authors a 2006 Interspeech paper with Hugo Van hamme on missing-data speech recognition. His doctorate in electrical engineering comes from KU Leuven.

  2. 2013

    Robust voice activity detection

    He co-authors a 2013 Interspeech paper on voice activity detection with Andreas Tsiartas and Shrikanth Narayanan. He completes postdoctoral research at the University of Southern California.

  3. 2020

    The DiPCo dinner-party corpus

    He co-authors the DiPCo dinner-party corpus at Interspeech 2020, a speech corpus recorded from groups of four Amazon employee volunteers. Before joining NVIDIA, he drives research efforts at Amazon across AWS and Alexa AI.

  4. 2023

    Predictive ASR for voice assistants

    In 2023 he co-authors Personalized Predictive ASR for Latency Reduction in Voice Assistants, a paper that Amazon Science lists as a publication.

  5. December 2025

    Nemotron 3 Nano paper

    In December 2025 he is an author of the Nemotron 3 Nano paper on arXiv. He also writes an NVIDIA Technical Blog post on building privacy-preserving evaluation benchmarks with synthetic data.

  6. September 2026

    NeMo Data Designer framework

    In September 2026 he is an author of NeMo Data Designer, an open-source framework for multi-modal synthetic data generation.

  7. As of October 2026

    Director of research at NVIDIA

    As of October 2026, he is director of research at NVIDIA, where he leads initiatives in synthetic data, privacy, and trustworthy AI. His previous role is VP of applied science at Gretel.

Sources and credits

Photo: We have no published photo of Maarten Van Segbroeck yet. Send us a free-licensed photo.

  1. ISCA Archive - Handling convolutional noise in missing data automatic speech recognition doi.org. Checked 15 October 2026.
  2. Author: Maarten Van Segbroeck | NVIDIA Technical Blog developer.nvidia.com. Checked 15 October 2026.
  3. ISCA Archive - A robust frontend for VAD: exploiting contextual, discriminative and spectral cues of human voice doi.org. Checked 15 October 2026.
  4. ISCA Archive - DiPCo — Dinner Party Corpus doi.org. Checked 15 October 2026.
  5. ISCA Archive - Personalized Predictive ASR for Latency Reduction in Voice Assistants doi.org. Checked 15 October 2026.
  6. Personalized predictive ASR for latency reduction in voice assistants - Amazon Science amazon.science. Checked 15 October 2026.
  7. [2512.20848] Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning arxiv.org. Checked 15 October 2026.
  8. How to Build Privacy-Preserving Evaluation Benchmarks with Synthetic Data | NVIDIA Technical Blog developer.nvidia.com. Checked 15 October 2026.
  9. [2609.17699] NeMo Data Designer: An Extensible Framework for Multimodal Synthetic Data Generation arxiv.org. Checked 15 October 2026.