Peter Kocsis

Peter joined NVIDIA as Research Scientist in 2026 April. His research focuses on inverse and forward rendering using generative image priors to solve fundamental challenges in material estimation and light transport. He finished his PhD at the Technical University of Munich, under the supervision of Prof. Dr. Matthias Nießner. Peter began his career as a mechatronics engineer. His early works span a broad domain from neural control for robotics to active learning for computer vision.

Nicolas Roussel

Nicolas Roussel is a research engineer at NVIDIA Research working with the Foundations of Graphics, Communications, and Machine Learning team. Prior to joining NVIDIA, he worked at EPFL, where he developed and maintained the Mitsuba 3 differentiable renderer, as well as its just-in-time compiler and automatic differentiation system Dr.Jit. He holds a M.Sc. in Communication Systems from EPFL. Outside of work, Nicolas enjoys rock climbing and cycling.

Bing Xu

Bing is a research scientist in the Real-Time Graphics Research Group at NVIDIA, where her work focuses on physically-based rendering and neural rendering. In the past, she explored forward and inverse light transport, importance sampling, denoising, and deep learning approaches for appearance modeling.

Bing earned her PhD from the University of California, San Diego, advised by Prof. Ravi Ramamoorthi. She holds a Bachelor’s degree from the University of Hong Kong and gained industry experience in offline rendering and 3D designer tools before moving into research.

Haozhe Liu

Haozhe Liu is a Research Scientist at NVIDIA Research. He received his Ph.D. from King Abdullah University of Science and Technology (KAUST), where he was advised by Jürgen Schmidhuber. Prior to joining NVIDIA, he worked at Meta AI (London), Meta AI (MPK), and Tencent.

QCalEval: Benchmarking Vision-Language Models for Quantum Calibration Plot Understanding

Quantum computing calibration depends on interpreting experimental data, and calibration plots provide the most universal human-readable representation for this task, yet no systematic evaluation exists of how well vision-language models (VLMs) interpret them. We introduce QCalEval, the first VLM benchmark for quantum calibration plots: 243 samples across 87 scenario types from 22 experiment families, spanning superconducting qubits and neutral atoms, evaluated on six question types in both zero-shot and in-context learning settings.

Fast AI-Based Pre-Decoders for Surface Codes

Fast, scalable decoding architectures that operate in a block-wise parallel fashion across space and time are essential for real-time fault-tolerant quantum computing. We introduce a scalable AI-based pre-decoder for the surface code that performs local, parallel error correction at low latency, removing the majority of physical errors before passing residual syndromes to a downstream global decoder. This modular architecture is backend-agnostic and composes with arbitrary global decoding algorithms designed for surface codes, and our implementation is completely open source.