Nolan Bryson

Full Stack · Game Dev · Graphics

About Me

I'm a full stack developer with a deep fascination for real-time graphics, 3D rendering, and the systems that power them.

I build graphics pipelines from scratch, work with Gaussian Splatting and novel view synthesis, and I love solving the hard problems that sit at the intersection of software and visual computing. I also dig into research papers and implement cutting-edge techniques in the field.

I enjoy problem solving, finding unique solutions, and designing systems to apply them. I optimize for performance and clarity in all stack layers. When I learn something new, I learn it from the very root and develop fundamental understanding before applying it to real-world problems. I thrive on analogies, mental models, and visualizing concepts. That's what draws me to graphics, it's an opportunity to turn math into something you can actually see and explore.

Currently Working On

Building Scout

A research pipeline application that lets users create custom scraping feeds for academic papers or any media outlet. Built algorithms that evaluate the importance of resources on a per-paper basis, helping users prioritize their time.

Open Source PlayCanvas

Contributed a bug fix to PlayCanvas React custom asset loading and a feature addition to the splat-transform pipeline. PRs merged upstream.

Work

Gaussian Splat Viewer

A real-time 3D Gaussian Splatting viewer built from the ground up. Renders millions of splats at interactive framerates using custom WebGL pipelines and compute-based sorting.

WebGL 3DGS Real-time Rendering

GIS Splat Interface

A geospatial interface for browsing and inspecting Gaussian Splat captures pinned to real-world coordinates. Navigate a globe, click a location, and dive into the scene.

GIS Geospatial 3DGS

Click anywhere to explore

Projects

I like picking up new tools and working in unfamiliar territory; most of what I know started as something I hadn't tried before. These projects span across different languages, pracitices, and concepts, and each one taught me something I wouldn't have learned staying in my comfort zone. I'm always continuously learning and exploring new technologies. I find fulfillment in picking up new skills and broadening my perspective in the field as a whole. Currently I'm trying to pick up C# for primarily game development, but I'm also learning about the .NET ecosystem and how to use it to build web applications. EntityFramework is pretty sweet and I've been very interested in the C# compiler and how it works.

Modor

A game engine built from scratch in C++ with Vulkan for rendering and CMake as the build system. Focuses on understanding every layer of the graphics pipeline from memory allocation to draw calls.

C++ Vulkan CMake Game Engine In Progress
View on GitHub →

engine.cpp

// Vulkan render pass setup
VkRenderPassBeginInfo renderPassInfo{};
renderPassInfo.sType =
    VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO;
renderPassInfo.renderPass  = renderPass;
renderPassInfo.framebuffer = framebuffers[imageIndex];
renderPassInfo.renderArea.offset = {0, 0};
renderPassInfo.renderArea.extent = swapExtent;

VkClearValue clearColor = {{{0.01f, 0.01f, 0.01f, 1.0f}}};
renderPassInfo.clearValueCount = 1;
renderPassInfo.pClearValues    = &clearColor;

vkCmdBeginRenderPass(
    cmdBuffer, &renderPassInfo,
    VK_SUBPASS_CONTENTS_INLINE);

Raytracer

A recursive raytracer built in Python as an academic project. Implements reflection, refraction, shadows, and multiple material types from scratch — no graphics libraries, just math.

Python Ray Tracing Computer Graphics
View on GitHub →

raytracer.py

def intersect_sphere(origin, direction, sphere):
    oc = origin - sphere.center
    a  = dot(direction, direction)
    b  = 2.0 * dot(oc, direction)
    c  = dot(oc, oc) - sphere.radius ** 2

    discriminant = b * b - 4 * a * c

    if discriminant < 0:
        return None

    t = (-b - sqrt(discriminant)) / (2 * a)
    if t > 0.001:
        hit_point = origin + t * direction
        normal = normalize(hit_point - sphere.center)
        return Hit(t, hit_point, normal)

AI/ML Implementations

A collection built during a Microsoft AI/ML Engineering certificate — hundreds of implementations spanning classification, NLP, computer vision, and deployed models.

Python PyTorch TensorFlow Machine Learning Deployment
View on GitHub →

training.log

$ python train.py --model resnet50 --epochs 30

Epoch [1/30]  loss: 2.4518  acc: 0.1247  lr: 1e-3
Epoch [5/30]  loss: 0.8931  acc: 0.6812  lr: 1e-3
Epoch [10/30] loss: 0.3472  acc: 0.8734  lr: 5e-4
Epoch [20/30] loss: 0.1205  acc: 0.9561  lr: 1e-4
Epoch [30/30] loss: 0.0847  acc: 0.9723  lr: 1e-5

Best checkpoint saved → model_best.pth
Val accuracy:  0.9681
Test accuracy: 0.9654

Deploying to Azure ML endpoint...  done

Contact Me!

Nolan Bryson

No-LAN-B · he/him

3rd Year CS Student developing passion projects on the side :)

Send me a message

For Game Developers

↑↑↓↓←→←→BA