Brady Zhou

bradybot-trap.no-botszhou (at) utexas.nospam.edu

I'm interested in anything computer vision / deep learning related to autonomous driving. I've worked on end-to-end driving from visual input, trajectory prediction, predicting maps from sensor data, 3D/4D neural scene representation, learned map representations. As of 2025, I'm finishing up my Ph.D. in Computer Science at The University of Texas at Austin, advised by Philipp Krähenbühl.

I've been at UT for quite a while - Hook em'🤘:

PhD Computer Science at UT Austin, 2019 -
BS+MS Computer Science at UT Austin, 2013 - 2018
BS Mathematics at UT Austin, 2013 - 2016

Outside of UT, I've been lucky to work with a bunch of smart people:


AI Resident @ Intel 2018 - 2019	Research Intern @ Wayve 2021	Research Intern @ Motional 2022	Research Intern @ NVIDIA 2022 - 2023

Research

Recent Work on Map Priors

Brady Zhou, Philipp Krähenbühl

Under Review

A framework for learning global spatially anchored features that help downstream perception.

Cross-view Transformers for Real-Time Map-view Semantic Segmentation

Brady Zhou, Philipp Krähenbühl

CVPR 2022, Oral

Enhance image positional embeddings with camera extrinsics for multi-view models.

code

Domain Adaptation through Task Distillation

Brady Zhou, Nimit Kalra, Philipp Krähenbühl

ECCV 2020

Sim2Sim driving by distilling a teacher that uses depth/segmentation as input.

code

Learning by Cheating

Dian Chen, Brady Zhou, Vladlen Koltun, Philipp Krähenbühl

CoRL 2019, Spotlight

code

A model trained robustly with map input provides good supervision for an image-only model.

Does computer vision matter for action?

Brady Zhou, Philipp Krähenbühl, Vladlen Koltun

Science Robotics 2019

Going through intermediate representations (depth, segmentation) helps agents act.

Science code

Don't let your Discriminator be fooled

Brady Zhou, Philipp Krähenbühl

ICLR 2019

Training GANs with adversarial examples smooths the loss landscape and improves generation.

OpenReview Poster code

GPU accelerated k-nearest neighbor kernel for sparse feature datasets

Brady Zhou, George Biros

UT MS Research

Accelerated KNN for sparse features by writing some fancy cuda kernels (bitonic sort).

link code

Teaching

I enjoy teaching!

Deep Learning: Online AI MS 2024 (1000+ students), Teaching Lead; handle 15+ TAs and handle grading server.
Deep Learning: Online AI MS 2023 (700 students), Teaching Lead; modernize course + materials.
Deep Learning: Online AI MS 2022 (400 students), Teaching Lead
Neural Networks, Fall 2020 (200 students), Instructor; lectures
Neural Networks, Fall 2019 (100 students), Teaching Assistant
Introduction to Programming, Fall 2015 (100 students), Teaching Assistant

Fun

Outside of work you will find me

Hanging out with my cats Pig and Teddy
At the bouldering gym
Tuning 3D printers
Building and flying FPV drones
Tweaking my tmux, VSCode, Vim configs + plugins
Playing with / coding up tools for LLMs