Machine Learning Systems Engineer Intern – 3D Generative AI

Job Description

Machine Learning Systems Intern | MeshyAI

The Tone:
This is a full-time internship at MeshyAI, located in Sunnyvale, California. The company builds a complete pipeline for 3D content creation and processing, encompassing text/image to 3D conversion, texturing, editing, and animation rigging. MeshyAI aims to develop the world’s largest end-to-end 3D native machine learning systems to unleash creativity for millions of users and enterprises. This role is crucial for building and scaling the core ML framework dedicated to 3D, from pretraining to inference.

The TL;DR
• Role: Internship
• Type: Full-time, 12 weeks or longer
• Location: In-person, Sunnyvale, California
• Pay: $50–$70 hourly
• Team: Collaborates with Machine Learning engineers
• Mission: Build and scale the end-to-end ML framework for 3D content generation and processing.
• Tech Stack: PyTorch, JAX, Python, C++, torch.compile, fully_shard (FSDP2) APIs, Triton kernels, bf16, fp8

What You’ll Actually Do
• Develop: Help build the end-to-end ML framework dedicated to 3D, covering pretraining, finetuning, and inferencing.
• Optimize: Scale high-throughput 3D data pipelines for foundational training and develop efficient inference engines for diffusion models.
• Engineer: Contribute to the design and implementation of training frameworks and novel model architectures.
• Maintain: Debug and monitor hardware platforms to ensure optimal performance of AI systems.
• Research: Collaborate with machine learning engineers to drive progress in AI and graphics R&D.

The Must-Haves
• Background: Undergraduate, master’s, or PhD student with strong technical fundamentals, intending to join full-time after graduation (ideally 9/2026-9/2027).
• Experience: Practical understanding of machine learning or high performance graphics, coupled with hands-on experience in machine learning or GPU development through research or engineering work. Solid practical understanding of at least one machine learning framework (e.g., PyTorch, JAX).
• Skills: Ability to write maintainable code in Python and/or C++, learn new concepts quickly, delve into complex codebases, and demonstrate a performance and efficiency-oriented mindset with attention to detail. Strong communication skills for a globally distributed team.
• Bonus: Experience navigating PyTorch internals (e.g., torch.compile, fully_shard (FSDP2) APIs), building Triton kernels, large-scale distributed training (e.g., DP, TP, CP, PP, zero redundancy optimizers), diffusion models in 3D or video, or low precision bf16 or fp8 training. First-author publications in top-tier conferences such as SIGGRAPH or SIGKDD are also a strong plus.