Gaussian Splat Creator

Summary Gaussian Splatting Pipeline Engineer

We have built a working platform that turns images and rendered video into interactive 3D Gaussian Splatting experiences for real estate.

Our system can create splats using third-party models, edit and segment scenes, remove surrounding buildings, reveal individual floors and floor plans, add unit labels, and publish the finished experience to the web.

We are looking for someone to take ownership of this existing pipeline. You will work with our lead programmer, learn the current codebase, produce high-quality splats, solve reconstruction problems, and continually make the system faster, more reliable, and easier for our production team to use.

What You’ll Do

• Turn Blender renders, video, and photographs into high-quality Gaussian Splats.

• Prepare and process source imagery for reliable reconstruction.

• Diagnose and correct problems such as poor camera alignment, noise, floaters, missing geometry, inconsistent color, reflective surfaces, and distorted results.

• Operate, maintain, and improve our existing Gaussian Splatting pipeline.

• Integrate better third-party GS models and tools as the technology evolves.

• Improve automation, output quality, processing speed, and ease of use.

• Maintain the tools that let us edit, segment, label, and publish interactive splat-based scenes.

• Help our artists and production team use the system successfully.

• Work within and improve an existing TypeScript and Python codebase.

What We Need

• Hands-on experience creating and troubleshooting 3D Gaussian Splats, NeRFs, photogrammetry, or similar image-based 3D reconstructions.

• Strong understanding of image capture, camera poses, 3D geometry, and reconstruction workflows.

• Experience with tools such as COLMAP, OpenCV, PyTorch, or comparable systems.

• Ability to inspect and improve reconstructed scenes rather than simply run a model with default settings.

• Professional experience with Python and enough TypeScript to work effectively in our existing application.

• Experience running GPU-based reconstruction or training jobs locally or in the cloud.

• Ability to understand a complex existing codebase and take increasing ownership of it.

• Strong debugging skills and a practical, results-oriented approach.

Helpful but Not Required

• Blender experience, particularly rendering video or image sequences for reconstruction.

• Experience with web-based 3D viewers, WebGL, WebGPU, Three.js, shaders, or point-cloud rendering.

• Experience optimizing, cropping, compressing, converting, or streaming Gaussian Splat or point-cloud data.

• Docker and cloud GPU experience.

• Experience building production tools for artists or other nontechnical users.

What Success Looks Like

You can independently take source images or Blender output and produce a strong interactive Gaussian Splat. When the result is poor, you can determine whether the problem comes from the source imagery, camera alignment, reconstruction model, processing pipeline, or viewer—and fix it.

• Over time, you will make our GS production pipeline more automated, more dependable, higher quality, and substantially easier to operate.