David Luebke

David Luebke

Vice President of Research, NVIDIA · New Experiences Research

25,000+citations60h-index250+papers, chapters & patents85+patents issued20years at NVIDIA ResearchGoogle Scholar, Nov 2026

Short bio

I helped found NVIDIA Research in 2006 after eight years on the faculty of the University of Virginia. Our work has shipped in NVIDIA products, helped shape the industry, and influenced future research directions in graphics, VR, AI, & wireless.

Today I lead New Experiences Research at NVIDIA (NXP for short), a group working on visual generative AI, human–AI interaction, and foundations of machine learning. We study fundamentals of math, algorithms, and human perception & performance (the "human in the loop"), across topics ranging from real-time graphics to immersive displays to wireless communications.

Selected honors

  • 2026SIGGRAPH Asia Test of Time AwardTowards Foveated Rendering for Gaze-Tracked Virtual Reality (2016)
  • 2025SIGGRAPH Asia Test of Time AwardNear-Eye Light Field Displays (2013)
  • 2022IEEE VR AcademyInaugural class
  • 2021IEEE VR Technical Achievement Award
  • 2016Fellow of the IEEEFor contributions to GPU computing and computer graphics
  • 2018–19Best of Show, SIGGRAPH Emerging TechnologiesTwo consecutive years
  • 2010–18Best Paper awardsIEEE VR 2017 · ACM SAP 2018 · ACM I3D 2010
  • 2008NVIDIA Distinguished Inventor
  • 2005ACM I3D Test of Time AwardPortals and Mirrors (1995)
  • 2001NSF CAREER Award

NXP Research Themes

Click a theme to expand
1995200020052010201520202026
Level of detail & perceptual graphics1995 – 2005

Real-time computer graphics is the art of doing less work without anyone noticing. During my graduate and early academic career at UNC and then Virginia, I worked on visibility, view-dependent mesh simplification, and driving level of detail with models of human perception; much of it is collected in the book Level of Detail for 3D Graphics. These principles transcend the polygon-centric graphics systems of that early work; for example foveated rendering incorporates gaze direction into rendering only what the visual system can perceive, and has resurfaced throughout my career in settings from ray tracing to display technology to shading archiectures and generative AI in modern games.

Level of Detail for 3D Graphics (book) · Virtual Monticello museum exhibit

GPU computing & foundations of machine learning2003 – 2008

In the early 2000s graphics processors became programmable enough to do other work. My group at UVA helped pioneer the emerging concept of GPU computing, using graphics hardware for numerical solvers and building simulators to study GPU architectures. After joining NVIDIA I helped evangelize GPU computing across academic research, and together with my colleagues I authored multiple surveys of the rapidly-developing field. Those surveys remain my most-cited papers. Along the way I co-taught an "Introduction to Parallel Computing" online class using GPU computing, eventually reaching over 100,000 students.

Today my team continues to research the technology underlying GPU computing. We currently focus on mathematical and algorithmic foundations of AI, such as autoguidance of diffusion models, or new approaches to quantization & sparsity in transformers.

GPGPU surveys · Udacity CS344, Intro to Parallel Programming

Selected Team Research
  • Guiding a Diffusion Model with a Bad Version of ItselfKarras, Aittala, Kynkäänniemi, Lehtinen, Aila, Laine · NeurIPS 2022
  • Nemotron-Flash: Towards Latency-Optimal Hybrid Small Language ModelsFu, Dong, Diao, Van keirsbilck, Ye, Byeon, Karnati, Liebenwein, Zhang, Binder, Khadkevich, Keller, Kautz, Lin, Molchanov
GPU ray tracing2006 – 2018

I helped found NVIDIA Research in 2006, and we quickly decided that making GPUs good at ray tracing would be a defining research agenda for our new group. Our early work on algorithms led to the NVIDIA OptiX platform which targeted professional graphics like CAD and film. In 2013 I kicked off a large cross-functional effort across NVIDIA that brought real-time hardware ray tracing to games in 2018 with the NVIDIA GeForce RTX, changing the landscape of graphics and gaming.

OptiX · RTX

VR/AR displays & human performance2013 – 2020

When VR returned to commercial relevance in 2013, I quickly decided that the display, not the rendering, was now the weak link. To invent a display that sat 15mm from your eye (far too close to be in focus), spanned over 180º field of view, supported the resolution and color discrimination levels of the human retina, and was as vanishingly unobtrusive as a pair of glasses--this was not simply a matter of optical engineering but a broad long-term cross-displinary research agenda. Starting with Near-Eye Light Field Displays, I built a research campaign to co-design optics, display technology, and rendering for human perception. Our work spanned light-field, varifocal, and holographic displays, foveated rendering, and gaze tracking. This is my most academically decorated research, with multiple best-paper and test-of-time awards; but as XR diminished in importance to NVIDIA, we wound down wearable display work and applied what we had learned about human performance to topics such as latency in competitive gaming.

SIGGRAPH Asia Test of Time Awards · IEEE VR Technical Achievement Award · NVIDIA Reflex

Generative AI, authenticity & human–AI interaction2018 – present

Researchers in my group did seminal work on applying neural networks to image synthesis. Our Instant Neural Graphics Primitives (a.k.a. InstantNGP) transformed the emerging computer vision subfield of neural radiance fields by making training interactive and rendering real-time. Our early work on neural denoisers for ray-traced graphics grew into DLSS, a major NVIDIA product that greatly increases the speed and quality of computer game rendering. Meanwhile, the same group invented StyleGAN, which I consider "the first generative AI that worked". StyleGAN generated photorealistic high-resolution images of human faces in milliseconds, a stunning "Holy Grail" moment that also brought a new awareness of dangers implicit in visual generative AI. We continue to work on the topic, including foundational studies of diffusion models, but I also launched an effort to study the risks alongside the promise. This effort led to participation in a large DARPA project on media forensics, and on broader aspects of content authenticity such as how to detect synthetic videos or verify identity in AI-mediated video calls. Our current research brings all this expertise to the question of how people and AI interact: understanding human intent through gaze, attention, and nonverbal cues to improve interaction with AI agents, avatars, and robots.

StyleGAN line · EDM · avatar fingerprinting · AI-mediated 3D video conferencing

Selected Team Research
  • A Style-Based Generator Architecture for Generative Adversarial Networks (StyleGAN)Karras, Laine, Aila · CVPR 2019
  • Elucidating the Design Space of Diffusion-Based Generative Models (EDM)Karras, Aittala, Aila, Laine · NeurIPS 2022
  • Instant Neural Graphics Primitives with a Multiresolution Hash EncodingMüller, Evans, Schied, Keller · SIGGRAPH 2022
newest firstcitations: Google Scholar, Nov 2026

About

David Luebke helped found NVIDIA Research in 2006 after eight years on the faculty of the University of Virginia. Today he is Vice President of Research at NVIDIA, where he leads a group of over 40 researchers called New Experiences Research. He received his Ph.D. in Computer Science under Fred Brooks at the University of North Carolina and a B.A. in Chemistry from Colorado College.

New Experience Research currently focuses on visual generative AI, on human-AI (and human-robot) interaction, and on the foundations of machine learning, human perception, and wireless communication. Luebke's personal research spans generative AI, real-time graphics, virtual and augmented reality, display technology, media forensics, and human perception. He is a Fellow of the IEEE and an inaugural member of the IEEE VR Academy; he also served on the Computer Science & Telecommunications Board of the National Academies. His honors include the IEEE VR Technical Achievement Award, Multiple Test of Time Awards from SIGGRAPH Asia (2025 and 2026) and ACM I3D (2005), and the NVIDIA Distinguished Inventor award. He has co-authored a book, a major museum exhibit, and more than two hundred papers, articles, chapters, and patents.

Personal interests

Outside work, David enjoys hanging out with his family, along with playing and coaching ultimate frisbee - especially when he gets to combine all three. He also enjoys playing racquetball and vibe-coding web apps.

EducationPh.D. Computer Science, UNC Chapel Hill, 1998 (advisor Fred Brooks) · B.A. Chemistry, Colorado College
ServiceBoard Member: National Academies CSTB · General Chair: High Performance Graphics 2016 and 2009, Interactive Ray Tracing 2008, Graphics Hardware 2007 and 2005 · Papers Chair: High Performance Graphics 2011, Graphics Hardware 2008, ACM I3D 2005
TeachingUniversity of Virginia, 1998–2006 · Udacity CS344, Intro to Parallel Programming
PatentsOver 100 US applications, more than 85 issued