Ohao Engine — PBR helmet render with image-based lighting
Shaoxuan · Frank · Yin Ohao Engine — PBR · IBL · ray tracing

Graphics &
Engine Engineer

Vulkan · CUDA · Neural Rendering · Simulation

I build the infrastructure that makes graphics possible — renderers, engines, and the math underneath.

MSc Computer Science, Linköping University (Graphics & Visualization). Most of my work sits at the intersection of GPU programming, rendering pipelines, and systems that other people build on top of. Currently a physics simulation & rendering intern at Freedo Technology; open to full-time engineering roles.

37,184 Lines · Vulkan Engine Standalone C++/GLSL renderer with deferred + hardware ray-tracing paths.
11 dB PSNR · Thesis Finding 3D Gaussian Splatting gain over NeRF on reflective surfaces (MSc thesis result).
209 GB Public Dataset 12-camera synchronized capture released for NeRF / 3DGS research.
40 FPS CUDA Path Tracer Monte Carlo with BVH on RTX 4060 at 1280×720, Russian roulette termination.
Featured Work
Cornell box, multi-light scene rendered with multiple importance sampling Cornell · Multiple Importance Sampling
ReSTIR direct lighting with 12 light sources ReSTIR · 12 lights, low variance
Cornell box rendered at 16 samples per pixel and denoised with Intel OIDN OIDN · 16 spp denoised
Mirror turntable · RT reflections

Ohao Engine

C++20 · Vulkan · GLSL · ~37K lines

A standalone Vulkan engine I wrote from scratch. Hybrid renderer with a deferred G-buffer path and a hardware ray-tracing path sharing the same scene representation, bindless resources, and acceleration-structure workflow.

Implements Cook-Torrance GGX BRDF, cascaded shadow maps, TAA, SSAO, and screen-space reflections on the deferred side; RT shadows and RT reflections via Vulkan KHR ray-tracing on the other. Physics layer uses GJK/EPA collision detection over a BVH.

Deliberately built without an editor or third-party engine wrappers — the goal was to understand every system end-to-end: render dispatch, scene upload, acceleration-structure build/update, offscreen image validation, the whole pipeline.

GitHub Repo ↗
Chess scene rendered in Ohao under studio HDR environment Chess · path traced · studio HDR
Same chess scene rendered via the deferred pipeline Chess · deferred · real-time
Lantern with emissive glass path traced Lantern · emissive + glass · PT
Lantern · turntable · RT reflections

Rendered in Ohao — Same Engine, Two Pipelines

glTF Sample Assets · 4K · 256 spp · OIDN

The chess scene (33 materials, 2K PBR textures) runs through both the offline path-traced pipeline (Cook-Torrance GGX, NEE, MIS) and the deferred pipeline (G-buffer, cascaded shadow maps, SSAO), sharing the same bindless texture system and scene upload path.

Lantern tests emissive materials as light sources — the glass casing acts as a volumetric-ish participant, the inner emissive element drives the scene illumination, and the turntable sequence validates ray-traced reflection continuity across camera motion.

Each path-traced still was rendered at 4K, 256 spp, denoised with Intel OIDN. End-to-end render time: ~50 seconds per frame on RTX 4060.

glTF Sample Assets ↗
12-camera capture rig setup Objects captured for the dataset Aerial reconstruction of Gränsö Castle

Neural Rendering Dataset — MSc Thesis

C++ · C# · WinUI · Hardware Control · ~8K lines

Built CamMatrixCapture: a 12-camera synchronized capture system with staggered trigger firmware, Bluetooth-controlled turntable automation, and a state-machine driven capture workflow. Indoor object captures feed NeRF / 3DGS training; outdoor work includes drone reconstruction of Gränsö Castle.

Released a 209 GB public dataset focused on challenging materials — reflective, transparent, translucent surfaces that traditionally break neural rendering methods.

The benchmarking contribution: 3D Gaussian Splatting reaches ~11 dB PSNR above NeRF on reflective surfaces — a surprisingly large gap, with clear implications for production capture pipelines.

Gränsö Reconstruction (video) ↗
Selected Projects

CUDA Path Tracer

C++ · CUDA · 2.8K lines

Monte Carlo path tracer with BVH acceleration, Russian roulette termination, multiple material models. 40 FPS on RTX 4060 at 1280×720.

Repo ↗

SPH Water Simulation

C++ · CUDA · OpenGL · 30K lines

GPU-accelerated SPH fluid with Marching Cubes surface reconstruction and CUDA/OpenGL interop for real-time visualization.

Chinese Chess AI

C++ · Qt · CUDA · 2.9K lines

Full Xiangqi game in C++/Qt with a Deep Q-Learning agent, CUDA-accelerated training, self-play, and model save/load.

Repo ↗
Experience
Freedo Technology
Physics Simulation & Rendering Intern · Remote
Feb 2026 — Present
  • Fluid simulation and differentiable rendering systems using Taichi, NVIDIA Warp, and nvdiffrast.
Independent Product Project
AI Platform Engineer · Part-time
2025 — Present
  • Built GPU-backed media generation workflows and async job orchestration for AI-assisted creative tooling.
  • Worked across frontend, backend, storage, auth, billing, and production-facing technical systems — shipped end to end rather than a single-surface contribution.
Beijing Guoyao Xintiandi Information Technology
C++ Development Engineer Intern · Beijing, China
Aug 2022 — Sep 2022
  • Built a Qt/OpenGL 3D visualization tool; improved rendering performance by 40% through profiling and optimization.
  • Implemented a custom C++ network protocol; improved data transmission efficiency by 50%.
Skills
LanguagesC++20, CUDA, Python, C#, TypeScript, Rust, GLSL, HLSL
Graphics APIsVulkan (incl. KHR ray tracing), OpenGL 4.3, DirectX 12
RenderingDeferred rendering, hardware RT, path tracing, PBR (Cook-Torrance GGX), global illumination, TAA, SSAO, SSR, CSM
GPU ComputeCUDA kernels, compute shaders, CUDA-OpenGL interop, Nsight / RenderDoc / PIX profiling
Simulation & MathSPH fluids, Marching Cubes, GJK/EPA collision, BVH, differentiable rendering
Neural RenderingNeRF, 3D Gaussian Splatting, synchronized multi-camera capture pipelines
Engines & ToolsCustom C++/Vulkan, Godot (GDExtension), Unreal Engine, Blender
SystemsCMake, Git, Linux, Docker, PostgreSQL, low-level performance profiling
How I Work

I care about infrastructure, not app surface area. The systems I find interesting are the ones other engineers build on top of: a renderer, a capture pipeline, a physics engine, a GPU scheduler. I profile before optimizing, I read the paper before reimplementing, and I'd rather ship one well-understood 37K-line engine than three shallow demos.

Education
Linköping University
MSc Computer Science — Graphics & Visualization
2023 — 2025
  • Graduated December 2025.
  • Top grades (5/5) in Advanced Global Illumination, Modelling & Animation, Computer Games Design, Large Distributed Projects.
  • Thesis: Neural Rendering Dataset Collection (see Featured Work above).
Beijing Information Science and Technology University
BEng Computer Science
2018 — 2023
  • GPA 3.53 · Advanced Mathematics 97 · CPU Design 91.
Publication

"Skiing Action Analysis Based on OpenPose"

Xu Z., Gao M., Yin S., Cui J.

Intelligent Computer and Applications, 2022