GPU Engineer Training Program 2025 (Full-time Salary)

Moreh is seeking engineers to join our team in developing AI software stacks for large-scale deep neural networks, cluster systems, and cloud infrastructure. Our framework is highly optimized for both training and inference across a variety of model architectures, such as large language models (e.g., DeepSeek, Llama) and multi-modal models (e.g., Stable Diffusion and Open-sora). Moreh's GPU software engineers tackle a variety of technical challenges to quickly run AI applications on different accelerator hardware, including GPUs and NPUs.  Implement and optimize various artificial neural network layers and tensor operations using OpenCL, CUDA, and other technologies Develop performance prediction and analysis models for tensor operations and different accelerators Create code generators and auto-tuners to generate and select the optimal kernel code for various GPU architectures Utilize GPU components specialized for AI applications, such as matrix multiplication units and high-speed interconnects between GPUs Develop efficient communication libraries for GPU-to-GPU communication, including communication between GPUs on different nodes.