Machine Learning Engineer (Training Optimization)

About the Role

Join our core generative AI team to push the boundaries of foundational model training, where you will design and optimize high-performance systems that power the next generation of creative intelligence.

Responsibilities

• Design, implement, and optimize large-scale distributed machine learning training systems.
• Improve GPU utilization, memory efficiency, and communication overhead for massive foundation models.
• Partner with research teams to align system architecture with complex algorithmic requirements.
• Profile and fine-tune training workflows, including the development of custom CUDA or Triton kernels.

Requirements

• Strong background in LLMs, multimodal AI, or diffusion models.
• Proficiency in Python and experience with distributed training frameworks like Megatron-LM, NeMo, or DeepSpeed.
• Deep knowledge of PyTorch or JAX and optimization techniques such as FSDP/ZeRO and gradient checkpointing.
• Hands-on experience writing custom GPU kernels in CUDA or Triton.

Nice-to-Haves

• Familiarity with system programming languages such as C++ or Rust.
• Experience in profiling and debugging large-scale GPU clusters.