Brennan Shacklett

I am a Ph.D candidate at Stanford University advised by Kayvon Fatahalian. My work revolves around designing graphics systems from the ground up for the new domain of training AI-based agents. Learning complex tasks requires agents to perform millions of simulated actions in virtual environments, presenting a challenging graphics workload with significant impact to research iteration times and accessibility. My current research focuses on evolving game engine and renderer architecture to enable high-throughput, GPU-accelerated simulation of thousands of learning environments simultaneously across a wide range of tasks. In the past, I've also worked on real-time image reconstruction techniques and debugging tools for hardware design. Prior to beginning my Ph.D., I received my B.S. and M.S. in Systems, also from Stanford University.

Publications

GPUDrive: Data-Driven, Multi-Agent Driving Simulation at 1 Million FPS

Saman Kazemkhani, Aarav Pandya, Daphne Cornelisse, Brennan Shacklett, Eugene Vinitsky

International Conference on Learning Representations (ICLR), 2025

Project Page | Paper

High-Throughput Batch Rendering for Embodied AI

Luc Guy Rosenzweig, Brennan Shacklett, Warren Xia, Kayvon Fatahalian

Proceedings of SIGGRAPH Asia 2024

Project Page | Paper | Code

Habitat Synthetic Scenes Dataset (HSSD):
An Analysis of 3D Scene Scale and Realism Tradeoffs for ObjectGoal Navigation

Mukul Khanna, Yongsen Mao, Hanxiao Jiang, Sanjay Haresh, Brennan Shacklett,
Dhruv Batra, Alexander Clegg, Eric Undersander, Angel Chang, Manolis Savva

IEEE / CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024

Project Page | Paper | Dataset | Code

An Extensible, Data-Oriented Architecture for High-Performance, Many-World Simulation

Brennan Shacklett, Luc Guy Rosenzweig, Zhiqiang Xie, Bidipta Sarkar, Andrew Szot, Erik Wijmans,
Vladlen Koltun, Dhruv Batra, Kayvon Fatahalian

Transactions on Graphics 2023 (Presented at SIGGRAPH 2023)

Project Page | Paper | Source code

Path Tracing of Terabyte-Scale Scenes using Thousands of Cloud CPUs

Sadjad Fouladi, Brennan Shacklett, Fait Poms, Arjun Arora, Alex Ozdemir, Deepti Raghavan,
Pat Hanrahan, Kayvon Fatahalian, Keith Winstein

Transactions on Graphics 2022 (Presented at SIGGRAPH 2022)

Project Page | Paper | Source code

Megaverse: Simulating Embodied Agents at One Million Experiences per Second

Aleksei Petrenko, Erik Wijmans, Brennan Shacklett, Vladlen Koltun

International Conference on Machine Learning (ICML), 2021

Project Page | Paper | Source code

Large Batch Simulation for Deep Reinforcement Learning

Brennan Shacklett, Erik Wijmans, Aleksei Petrenko, Manolis Savva, Dhruv Batra, Vladlen Koltun, Kayvon Fatahalian

International Conference on Learning Representations (ICLR), 2021

Project Page | arXiv | Training & Simulator Source code | Renderer Source Code

Encoding, Fast and Slow: Low-Latency Video Processing Using Thousands of Tiny Threads

Sadjad Fouladi, Riad Wahby, Brennan Shacklett, K Balasubramaniam, W Zeng, R Bhalerao, A Sivaraman, G Porter, and Keith Winstein

USENIX Symposium on Networked Systems Design and Implementation (NSDI), 2017

Paper | Source code

Projects

High Quality Rendering for Reinforcement Learning

Using modern raytracing APIs to build a physically based rendering framework designed from the ground up for reinforcement learning and other ML use cases.

Source Code: github.com/shacklettbp/rlpbr

Continuous Learning for Interactive Denoising of Path Traced Content

Masters Thesis, advised by Anjul Patney, Marco Salvi, Aaron Lefohn, and Kayvon Fatahalian

Source code: github.com/shacklettbp/live-denoiser

Interactive Circuit Debugging and Simulation with Just In Time Compilation

Undergraduate Honors Thesis, advised by Pat Hanrahan

Source code: github.com/shacklettbp/jitsim

Teaching & Industry Experience

6/23 - 12/24

Research Associate at Treyarch / Activision Central Technology

Exploring game development applications of rapid agent training.

6/21 - 9/21

Research Intern in the Intelligent Systems Lab at Intel

Working on 3D simulation and rendering for reinforcement learning and robotics under Vladlen Koltun.

4/19 - 6/19, 4/20 - 6/20

Head Course Assistant for CS348B: Image Synthesis Techniques

Developed new assignments and assisted students. Class taught by Pat Hanrahan and Matt Pharr.

6/19 - 9/19

Research Intern in the Real-Time Rendering group at NVIDIA Research

Worked with Marco Salvi on image supersampling techniques (DLSS) under Aaron Lefohn.

6/16 - 9/16

Research Intern in the Video Compression group at Mozilla Research

Improved performance of the AV1 rate-distortion optimizer under Tim Terriberry.