About the job
Who You Are
We are seeking talented Machine Learning Systems Engineers to contribute to the development of the world's largest end-to-end 3D native machine learning systems. You will collaborate on our comprehensive ML framework tailored for 3D applications, encompassing pretraining, fine-tuning, inference, and more. We value strong hands-on engineering skills, a passion for learning, and an ability to excel in a dynamic, high-responsibility environment.
Who We Are
At Meshy, we envision a world where 3D creation is limitless and accessible to all. Our mission is clear: unleash creativity. We have developed a comprehensive pipeline for 3D content that spans text/image to 3D, texturing, texture editing, animation rigging, and beyond. Additionally, we foster a vibrant community for creators to share their work, draw inspiration from others, and utilize our platform as an asset marketplace for their games and prototypes. Recognized as the No.1 in popularity among 3D AI tools (according to the 2024 A16Z Games survey), Meshy delivers real value to enterprises such as Meta, Square Enix, and DeepMind, as well as millions of end-users. Our technology powers game and film production, 3D printing, industrial product design, user-generated content features, and even training simulations for robotics and physical AI.
Your Next Challenge
3D is the exciting new frontier of Generative AI, and your role at Meshy will present unique challenges in both training and inference. You will engage with the full stack of AI, from debugging and monitoring hardware platforms, building training frameworks, scaling high-throughput 3D data pipelines, collaborating with researchers on novel model architectures, to developing efficient inference engines for diffusion models and more. Here are some specific challenges on the training side:
Collaborate closely with researchers to co-design the next frontier of 3D & Spatial AI.
Develop and refine modern PyTorch solutions for maximum parallelism and efficiency, establishing a clean and intuitive training infrastructure for our foundational models.
Identify bottlenecks and optimize for high throughput & efficient distributed model training across hundreds to thousands of GPUs.
Implement and maintain 3D-specific custom operators in Triton or CUDA.
Design and uphold novel data-loading frameworks and libraries.