AI Performance Engineer (Cloud AI Engineering), Sr | Staff | Sr. Staff - San Diego, California, United States of America

Originally posted 9 April 2026 by the employer.

Optimize and deploy LLM, VLM, and diffusion models for inference acceleration using PyTorch, ONNX, and Triton.

About the role

The Qualcomm Cloud AI team develops hardware and software solutions for inference acceleration across deep learning workloads. In this role, you will analyze, convert, and optimize models for efficient inference and map AI workloads to current and future hardware designs.

What you'll do

- Convert, optimize, and deploy models for efficient inference using PyTorch and ONNX under throughput and latency constraints. - Conduct performance analysis and optimization of LLM, VLM, and diffusion models, focusing on advanced algorithms such as attention mechanisms and MoEs. - Map AI workloads on top of current and future hardware designs using workload mapping strategies including sharding and various parallelisms. - Design and implement high-level kernels in Triton with a focus on generating efficient, low-level code. - Analyze complex performance or stability issues to identify the root cause of underlying problems. - Collaborate with internal compiler, firmware, and platform teams to drive customer solutions and deliver continuous insights into workload performance.

What you'll need

- Bachelor's degree with 6+ years, Master's degree with 5+ years, or PhD with 4+ years of Hardware Engineering, Software Engineering, Systems Engineering, or related work experience. - Hands-on experience building and optimizing language models in PyTorch and ONNX. - Strong Python programming skills. - Understanding of transformer architectures, attention mechanisms, and performance trade-offs. - Understanding of computer architecture, ML accelerators, in-memory processing, and distributed systems. - Experience with workload mapping strategies exhibiting sharding or parallelisms.

Nice to have

- Background in neural network operators and mathematical operations, including linear algebra and math libraries. - Understanding of machine learning compilers. - Experience converging accuracy and its evaluation methods. - Knowledge of torch.compile or torchDynamo. - PhD in Computer Science, Computer Engineering, or Machine Learning.

Skills: inference optimization, PyTorch, ONNX, LLM, VLM, diffusion models, hardware designs