Machine Learning System Engineer

Headquartered in Silicon Valley, Meshy is the leading 3D generative AI company on a mission to Unleash 3D Creativity by transforming the content creation pipeline. Meshy makes it effortless for both professional artists and hobbyists to create unique 3D assets—turning text and images into stunning 3D models in just minutes. What once took weeks and cost $1,000 now takes just 2 minutes and $1.

Our world-class team of top experts in computer graphics, AI, and art includes alumni from MIT, Stanford, and Berkeley, as well as veterans from Nvidia and Microsoft. Our talent spans the globe, with team members distributed across North America, Asia, and Oceania, fostering a diverse and innovative multi-regional culture focused on solving global 3D challenges. Meshy is trusted by top developers, backed by premiere venture capital firms like Sequoia and GGV, and has successfully raised $52 Million in funding.

Meshy is the market leader, recognized as the No.1 in popularity among 3D AI tools (according to 2024 A16Z Games) and No.1 in website traffic (according to SimilarWeb, with 3 Million monthly visits). The platform boasts over 5 Million users and has generated 40 Million models.

Founder and CEO Yuanming (Ethan) Hu earned his Ph.D. in graphics and AI from MIT, where he developed the acclaimed Taichi GPU programming language (27K stars on GitHub, used by 300+ institutes). His work is highly influential, including an honorable mention for the SIGGRAPH 2022 Outstanding Doctoral Dissertation Award and over 2,700 research citations.

About the Role

What You’ll Do:

• Work within the AI model team to streamline 3D data into high-throughput pipelines and scale training infrastructure to hundreds of GPUs.
• Train, accelerate, and deploy machine learning models for 3D GenAI.
• Design and implement reliable and scalable distributed training pipelines, optimize end-to-end training efficiency.
• Work closely with researchers, software engineers, and artists to integrate AI models into production.

• Work closely with researchers to build the training infrastructure for our in-house foundational models.
• Identifying bottlenecks and optimizing for high throughput & efficient distributed model training across hundreds to thousands of GPUs.
• Building and maintaining training clusters and job schedulers.
• Implementing and maintaining 3D specific custom operators in Triton or CUDA

• Building efficient inference endpoints with complex model pipelines
• Optimizing models through compilation, fusion, quantization, etc.

What We're looking for:

• Experience in machine learning or high performance graphics.
• Solid practical understanding of at least one machine learning framework (e.g. PyTorch, Flax).
• Strong ability to write beautiful and maintainable code in Python and/or C++.
• Ability to learn fast and dive into new concepts or complex codebases.
• Performance and efficiency oriented mindset, with a strong interest in the tiniest detail.
• Strong communication skills for working in a globally distributed team.

Nice to have:

• A strong passion to navigate through the PyTorch internals, with hands-on experience in areas like torch.compile , fully_shard (FSDP2) APIs.
• Experience with building Triton kernels.
• Experiences with large-scale distributed training, familiarity with modern parallelization techniques: DP, TP, CP, PP, zero redundancy optimizers, etc.
• Experience with diffusion models in 3D or video.
• Experience with full bf16 or partially fp8 training.