About


Hi! I'm Winnie, a research engineer at Netflix.

I graduated in 2022 with a Ph.D in Computer Science, advised by Professor Ron Fedkiw. My academic interests mainly lie at the intersection of Applied Mathematics, Visual Arts, and Machine Learning; this includes but is not limited to Graphics, 3D Vision, and Computational Geometry.

Summaries of my , , and experience can be found here, as well as a small collection of my . Please refer to my resume for additional information.

In my free time, I like to cook, doodle, sew, read fantasy/sci-fi, and belt (screech) out Taiwanese rock ballads on my commute.




Leveraging Deepfakes in Facial Capture Pipelines

arxiv preprint

Thesis work. Developed method for personalized appearance acquisition and facial capture via minimal unsupervised training on personalized data.


Autoencoders for Inverse Rendering

CS231N:Convolutional Neural Networks (2017)

Experimented with physics-based image feature extraction. Synthesized toy dataset of textures, depth, normals, and radiance by augmenting raytracers. Trained adversarial autoencoders for image generation.


Three Dimensional Reconstruction of Botanical Trees with Simulatable Geometry

SCA 2021

Co-author. Developed segmentation networks, segmentation-based advection algorithms, and user interfaces for assisted branch extraction from images.


Adaptive Samplers for Ray-Tracing

CS229:Machine Learning, best project in Vision (2015)

Implemented SVM-based hierarchical adaptive sampler in C++ and pbrt.


Real-time Interactive Tree Animation

IEEE TVCG.2017.2661308 (2017)

Co-author. Augmented in-house raytracer for visualizations of simulation results, scripted cluster computing and scene generation routined, verified the mathematical soundness of some parts of the algorithm.


Body Pose from Optical Flow

CS231A:Computer Vision (2015)

Experimented with optical flow based methods and Kalman filters for joint tracking in videos.


Deformation Graphs for Meshes

Undergraduate Research in Computer Science (2014)

Summer internship project in Geometric Computation Group. Implemented constraint-based deformations for use in research framework.


Deterministic Growth Processes

MATH101:Math Discovery Lab (2015)

Group project of 3. Collaborated with Caroline Ellison and Mark Nishimura to explore the mathematical properties of a well-defined growth process.


Trayl

CS248:Interactive Graphics, 1st place game (2015)

Unity, Blender, and Photoshop. Worked with Herman Chau to produce this game. Implemented custom acceleration structures, created assets, and designed UI. Click here for more details.


Procedurally Generated Ice Cream

CS348B:Image Synthesis Techniques (2014)

C++ and pbrt. Procedurally modeled soft serve ice cream, via implementation of shape classes in pbrt.


Double Pendulums

CS448:Scientific Computing (2015)

Visualized with C++ and OpenGL, arranged in Adobe Illustrator. Rendering and simulation of paths of double pendulums.


Underwater Scene

CS148:Introductory Graphics (2013)

C++ and class raytracer. Worked with Adam Young to produce this image. Extended raytracer to create underwater lighting via implementation of procedurally generated waves and Snell's law.




Computational Methods for Content Creation

Netflix 2023-

I am a Studio Algorithms research engineer, working on combining computer graphics and machine learning algorithms for use in visual content creation (ex. VFX and animation).


Facetracking for AR/VR

Facebook 2020, 2022-2023

I was a research intern in the People AI team at Facebook AR/VR, and I worked on designing, implementing, and analyzing a variety of deep-learning based face-tracking methods, with the goal of productizing promising methods to be used on-device for AR/VR applications. I joined the team as a full-time software engineer in 2022, and my main role was helping ship face tracking onto the Quest Pro virtual reality headset. I also helped maintain Oculus lipsync technology and prototyped artist-guided AI/ML facial rigs.


Deepfake-based face tracking

Epic Games 2021

I was an intern in the physics team at Epic Games, and worked on exploring the potential of utilizing deepfake technology in facial capture pipelines, as well as utilizing Unreal Engine and Metahuman capabilities for synthetic data generation.


VFX Research and Development

Industrial Light and Magic 2018-2020

I was a research intern at Industrial Light and Magic in summer of 2018, and continued working as a part-time research engineer up until March 2020. I worked on researching, implementing, and evaluating traditional and modern computer vision algorithms, with the goal of improving animation pipelines through tracking, retargeting, and generation of 2D and 3D data. More specifically, I've had the privilege to design and develop some data-driven production tools for The Irishman (2019) and Terminator: Dark Fate (2019).


Stonehearth Game Development

Radiant Entertainment 2015, Riot Games 2016

I was an software engineering intern at Radiant Entertainment in 2015, and an intern again in 2016 after Radiant Entertainment's acquisition by Riot Games. My first summer I worked on the restructuring, implementation, and asset design of their procedural world generation code, as discussed here, and in the following summer I worked on the restructuring and implementation of the in-game hydrology system.



Stanford Computer Science

I was co-instructor of CS148: Introduction to Computer Graphics in Summer 2022, and have been teaching assistants for

(** Denotes head TA positions, where I led teams of 5-25 teaching assistants depending on class size.)

High School Outreach

I wrote the curriculum, led development, and cotaught the graphics track for the Girls Teaching Girls To Code CodeCamp (2015,2016).
Curriculum was designed for high school girls with minimum coding experience.
Here is the handout and the source code for 2016, designed to teach students how to create a simple game in Unity.
I cotaught the GTGTC summer workshop (2014) on 3D printing,
and I have taught a few sessions on graphics and image processing at Stanford Splash, a program for middle school and high school students.

Mathematics for Advanced High School Students

I was a teaching assistant for Euler Circle in Abstract Algebra (2015) and Cryptography (2016),
and was a residential teaching assistant in Number Theory for Stanford University Mathematics Camp (2013).



In the past few years, some areas I've dabbled in include 2D/3D graphic design, sketching, sewing, photography, interactive exhibition design. Here is a small collection of results: