High-Performance Graphics Software Development

Supercharge your visually-demanding applications with next-generation rendering solutions, high-performance compute, and relentless optimization.

What our partners say

Phil Langrishe

Phil Langrishe

Managing Director at Applications in Cadd

Company Logo

Matt, Erfan and the rest of the Devsh team have been amazing to work with. We presented them with a very difficult challenge which was to take a large and very old, plus some what outdated C++ codebase written for a different time, and bring it up to date so that we could make use of their Vulkan based Nabla graphics engine. They broke the challenge down into manageable chunks and then over the past few years have proceeded to inject lightening fast graphics into our application. All agreed targets have been met on time, and on budget. I can not thank them enough for the hard work and dedication they have shown towards achieving our goals. Their knowledge and professionalism has been second to none!

Yoran Bosman

Yoran Bosman

Senior Software Architect

Company Logo

The DevSH team transformed our CPU render farm for architectural visualization into a scalable GPU-based system with outstanding performance and reliability. Their expertise and commitment to achieving the best results were clear throughout the project. They maintained full backward compatibility, ensuring a smooth transition with minimal disruption to our workflows.

Commercial work

Engineered Solutions for Our Partners

Interactive GPU Path Tracer

Ditt

Real-time and offline rendering work for interior design workflows using OpenCL, OptiX, Mitsuba and Vulkan.

  • Interactive GPU Path Tracer (OpenCL and OptiX) ingesting existing Mitsuba scenes
  • Maintenance of in-house interior design tool
  • Integration of new Real-Time rendering techniques for rapid previews in the design tool
  • Denoisers (AI, RWMC, etc.) and Image Processing (e.g. FFT Convolution Bloom)
  • Consulting w.r.t. existing Mitsuba render farm
  • Ongoing Vulkan Real-Time Path Tracer development

Lindt Office

GPU-Driven Graphics Engine for n4ce v5.0

Applications in CADD

A tailored GPU-driven renderer for large point cloud and civil engineering datasets on top of Nabla.

  • Development of a tailor-made, GPU-driven graphics engine for n4ce v5.0 using the Nabla platform
  • Delivered a 100x performance improvement over the legacy system, transforming rendering times from multi-second bottlenecks into fluid, real-time visualization
  • Engineered for extremely large point cloud datasets and modern large-scale civil engineering projects
  • Significant performance improvements over the legacy n4ce renderer, designed to remain scalable for the next years
  • Close collaboration to integrate the new renderer seamlessly into the n4ce v5.0 product
  • Design and implementation of a modern build system, including shader preprocessing & embedded SPIR-V tooling
  • Ongoing optimisation and support during the lead-up to the public BETA release

n4ce by Applications in CADD

n4ce by Applications in CADD

Vulkan Mobile GPU Rendering

Wild Software Inc.

Deep Vulkan work for mobile GPUs, focused on synchronization, bindless rendering and tile-based renderer constraints.

  • Solving Vulkan synchronization issues
  • Implementing bindless rendering
  • Work around DXC and Adreno 600 limitations to allow the usage of Buffer Device Address
  • Implementation of GPU-driven rendering
  • Occlusion culling with a novel algorithm specifically designed for mobile TBDR GPUs (see our Vulkanised 2026 talk)
  • And much more

Factions by Wild Software Inc.

GPGPU Computer Vision

Imverse

Computer vision acceleration work for real-time volumetric capture and RGB+D processing.

  • GPGPU Consulting and Contracting to solve Computer Vision problems
  • GPU Accelerated Silhouette Carving from RGB+D real-time video inputs

Architecture study

Web & Native Renderer Feasibility

Feasibility work around .NET, WASM, Blazor, TypeScript and C# interop for a unified renderer path.

.NET / WASMInteropRenderer architecture

Architecture study

RELEX Solutions

  • Investigation of the new .NET 6 WASM and Blazor SDKs (example issues reported to Microsoft)
  • Feasibility studies of TypeScript and C# interoperation and Unified Web & Native Renderer

Performance appraisal

Renderer Performance Analysis & Design

Performance diagnosis and renderer architecture planning for complex engineering scenes.

ProfilingBottlenecksRenderer design

Performance appraisal

Synera

  • Deep performance analysis identifying Rendering bottlenecks on complex scenes
  • Plan of action and initial design for a new Renderer
  • Conclusion and recommendations for the renderer redesign

Photogrammetry & Point Cloud Pipeline

Build A World Aps.

World reconstruction work spanning point clouds, LiDAR, OpenGL rendering, simulation and voxel physics.

  • Photogrammetry (generating point clouds from drone flythroughs)
  • LiDAR Point Cloud Processing and Volume Reconstruction (3D game worlds from aerial scans)
  • Maintaining and Extending the existing OpenGL engine
  • GLSL Shader Development (ocean simulation, particle simulation)
  • Distributed Networked Fluid Simulation
  • Physics Engine Development (high performance voxel connectivity for destruction)
Nabla

Nabla

Our Open Source Rendering Framework

Devsh-Graphics-Programming / Nabla
685

Stars

73

Forks

40

Open PRs

15,969

Commits

  • Single-source HLSL202x/C++20 Standard Template Header-Only Library
  • SPIR-V and Vulkan as First-Class Citizens
  • Utilities for Rapid Prototyping for modern GPU Workloads (raytracing, compute, raster, etc.)
  • Unit-Tested BxDFs for Physically Based Rendering
  • Intelligent GPU Object Lifecycle Tracking and Resource Management
Raytracing

Raytracing

Fluid Simulation

GPU-Accelerated Vectorized Linework

Nabla Shader Compiler & Godbolt docker integration

Nabla Shader Compiler & Godbolt docker integration

Fast Fourier Transform Bloom

ImGui Integration

ImGui Integration

2D Constructive Solid Geometry

Iridescent Materials

Iridescent Materials

Community and standards

Advancing the Ecosystem

We actively contribute to the graphics programming community through our technical blog and presentations at major conferences like Vulkanised and the Shading Languages Symposium.

Presenting at Shading Language Symposium 2026

Presenting at Shading Language Symposium 2026

Our booth at Vulkanised 2026

Our booth at Vulkanised 2026

Presenting at Vulkanised 2023

Presenting at Vulkanised 2023

Presenting at Vulkanised 2026

Presenting at Vulkanised 2026

Who We Are

DevSH Graphics Programming Sp. z O.O is a specialized collective of graphics engineers and mathematicians focused entirely on GPU architecture, rendering, and High-Performance Computing. We are not a traditional software house. We don't build CRUD apps, and we don't do repetitive contract work. Instead, our consultants integrate directly alongside our clients' engineering teams to architect renderers, simulations, and compilers at the frontier of what hardware can do.

DevSH has been a remote-first company since its inception, while building unparalleled credibility amongst GPU engineers thanks to how we innovate on every project, resulting in a constant backlog of applications to our company even when it involves a pay cut. We capture top-tier engineering talent, including PhDs, ex-IHV driver developers, and AAA rendering architects, who wouldn't even consider applying for a FAANG/MANGO job posting if it required relocation. Becoming our client lets you tap this pool of otherwise unreachable talent, sending a signal to everyone else that real engineering gets done at your company.

We invest heavily in our own R&D and Open Source middleware, most notably Nabla, our rapid prototyping framework designed to deliver a CUDA-like programming experience within the Vulkan ecosystem. By committing to open-source infrastructure, we eliminate boilerplate and ensure we never have to solve the same foundational problems twice.

Because we hire the absolute best wherever they live, our globally distributed team maintains regular working hours that overlap all the way from San Francisco to Sydney. We train and work with the elite. Our alumni have gone on to drive the industry forward at Intel, ARM, and Apple as core driver and dev-tech engineers, as well as architecting engines for AAA games.

⬅️a friendly froge➡️