I am a Machine Learning researcher at BAIR, advised by Yutong Bai and Alyosha Efros, and Founder of Buzzer Labs. I graduated from Physics and Computer Science at UC Berkeley.
I am interested in understanding intelligence and how it comes to be, how models scale, and what prevents them from doing so. I also have a fondness for design, interfaces, and hardware.
hi [at] jdiazchao [dot] comPreprint, 2026
TL;DR Standard bidirectional video diffusion breaks down as chains of dependent events lengthen. Its denoising loop can’t supply the scalable serial computation that simulation and reasoning tasks demand.
UIST, 2026
TL;DR We turn the physical world into a programming interface. Users teach agents to perform physical activities through narrated demonstrations in AR, turning actions and words into editable programs.
Preprint, 2024
TL;DR We repurpose diffusion attention to edit one image using another as a reference, without additional training or gradient-based guidance.
Turn the physical world into a digital interface. AI-powered smart glasses that learn your custom actions. Turn a lamp on with a tap of your desk, or pet a plushie to bring it to life. On-device AI for responsive, private workflows.
An AI-powered virtual wardrobe that understands your unique style, delivering personalized, context-aware outfit suggestions and actionable wardrobe insights for smarter shopping and sustainable fashion.
Life-changing. An advanced model rocket powered by four class D motors, with thrust-vector control, custom avionics, and integrated roll control and recovery systems. Guided by quaternion-based PID guidance software.
CS 280 — Graduate Computer Vision (Angjoo Kanazawa and Alyosha Efros)
Teaching Assistant, Spring 2026
CS 180 — Computer Vision and Computational Photography (Angjoo Kanazawa and Alyosha Efros)
Teaching Assistant, Fall 2025
