Dylan Hu

About me

I recently graduated from Brown University in 2024 with a B.S. in computer science and a B.A. in applied mathematics.

I work on Azure observability at Microsoft in Redmond, WA. Previously, I interned at Hume AI in New York, where I worked on the Empathic Voice Interface and internal tools for ML and AI research.

My primary research interests lie in 3D visual computing and how computer graphics and computer vision can enable interactive applications for capturing, augmenting, and understanding the world in 3D.

In my free time, I enjoy following competitive esports and brewing coffee.

Research

I am fortunate to have been a member of the Brown Visual Computing group. I was advised by Professor Srinath Sridhar as a research assistant in the Brown Interactive 3D Vision and Learning Lab.

Recently, I worked with Professor Daniel Ritchie on spatially-varying noise pattern generation with diffusion models.

Previously, I worked with Professor James Tompkin on neural factorization of 3D scenes through differentiable rendering. I have also investigated methods for segmentation in 3D scene representations using generative models and foundation models in segmentation.

Projects

EZNeRF is an educational implementation of NeRF written from scratch in PyTorch.

Carnot is a work-in-progress toy 3D game engine written in Rust and powered by WebGPU.

I have built a number of platforms and tools for the Brown University community:

Scenes is an interactive 3D scene viewer and editor implementing a custom JSON scenefile format built for Brown's introductory computer graphics course, CS1230.

Here! is a platform that enables TAs and students to manage course section registration and attendance.

Dear Blueno is a truly anonymous forum with over 6,000 threads and 1,000 users.

I have also implemented physically-based cloth simulation, a path tracer, and finite element simulation.