Description
Canva is hiring a Machine Learning Engineer to design, implement, and optimize large-scale distributed training systems for multimodal and foundation models. The role focuses on improving GPU utilization, communication overhead, and memory efficiency; partnering with research and modeling teams; and applying frameworks such as Megatron-LM, NVIDIA NeMo, FSDP, and Triton, including custom CUDA or Triton kernels. Candidates should have strong LLM, multimodal AI, or diffusion-model knowledge, Python proficiency, PyTorch or JAX expertise, and experience with distributed training optimization.
