Senior Machine Learning Engineer - GPU Acceleration

About the opportunity

Accelerate transformer training and multi-node inference pipelines using CUDA, TensorRT, and Megatron-LM.

What you will do

• check_circle Optimize deep neural network execution kernels using CUDA, Triton, and TensorRT.
• check_circle Implement pipeline, tensor, and data parallelism strategies for trillion-token pretraining runs.
• check_circle Benchmark and reduce inference latency for generative video and vision models on edge and data center GPUs.
• check_circle Engage with open-source AI frameworks (PyTorch, vLLM) to upstream acceleration features.

What we are looking for

• arrow_circle_right Bachelor’s or Master’s in Computer Science, Electrical Engineering, or related discipline.
• arrow_circle_right 5+ years hands-on experience training and profiling deep learning models on NVIDIA GPU clusters.
• arrow_circle_right Mastery of PyTorch internals, CUDA C/C++, and distributed communication libraries (NCCL).
• arrow_circle_right Deep understanding of transformer attention optimizations (FlashAttention, KV cache compression).

Skills & Tech Stack