About the role:

We're seeking a highly skilled AI/ML Research Engineer to join our founding team. The ideal candidate will be responsible for designing, implementing, optimizing various AI/ML models for 3D graphics, computer vision, robotics, and natural language understanding tasks, and deploying them on the web. This role requires deep technical expertise, a passion for working with 3D Gaussian Splatting (3DGS), and a strong startup mentality.

Our Gauzilla Pro is a fully web-based, AI-powered, and no-code editor/viewer for 3D/4D digital twins. It utilizes serverless edge AI models that run directly in web browsers for 3D object detection & segmentation, etc. As an AI/ML Research Engineer focused on product values, you’ll be helping us delivering AI-powered features on Gauzilla Pro. This is a unique opportunity where you'll influence strategic product and infrastructure decisions as we bridge the gap between 3DGS and the delightful experience our users have come to expect.

Key responsibilities:

• Work closely with the strongly hands-on CTO to design, develop, and maintain practical & robust AI/ML models for eg. 3D object segmentation, point cloud registration, visual/language recognition, etc.
• Collect, process, and store training datasets from various sources (including synthetically generate data), and build DataOps & MLOps pipelines for training, optimizing/finetuning models on the GPUs, and deploying them on the web.

Qualifications

• Bachelor's or Master's degree (prefered) in Computer Science (machine learning, computer graphics, computer vision), Engineering (robotics, SLAM), or a related field.
• 3+ years of experience in AI/ML model training and deployment for web platforms (WebAssembly in Rust/C++, ONNX with PyTorch/Tensorflow, WebGPU, WebNN).
• Strong understanding of ETL processes, data modeling/warehousing, and MLOps.
• Strong understanding of 3D Gaussian Splatting training/optimization on the GPU.

We strongly encourage candidates of all different backgrounds to apply. We believe that our work is stronger with a variety of perspectives.