Staff AI Model Optimization Architect for LLMs & Multimodal

Qualcomm; Austin (TX); On-site; USD 158,000 - 238,000; Full time

Job summary

Qualcomm Technologies, Inc. seeks a Staff Engineer to lead end-to-end AI model optimization for LLMs, VLMs, and diffusion on Qualcomm accelerators. You will transform PyTorch models, manage KVcache behavior, and drive deployment with PyTorch, ONNX, and torch.compile across multi-core systems.

You will collaborate with compiler and performance teams to craft lowering strategies and scalable tooling, ensuring high throughput with low latency and production-grade reliability.

Qualifications

• Bachelor's degree in CS/Engineering with 4+ years of related work experience.
• Master's degree with 3+ years of related work experience.
• PhD in CS/Engineering or related field with 2+ years of related work experience.
• Experience in hardware/software systems and ML accelerators preferred.

Responsibilities

• Architect and deliver model optimization strategies for efficient inference on Qualcomm accelerators.
• Drive graph capture and deployment using PyTorch, ONNX, and torch.compile.
• Design fusion kernels via DSLs (e.g., Triton) and perform kernel fusion.
• Collaborate with compiler, performance, and accuracy teams on lowering strategies.
• Profile and optimize LLM/VLM/diffusion inference for throughput and latency.
• Own transformer optimizations including KVcache management and decoding behavior.
• Support distributed inference strategies across multi-core/multi-device systems.
• Create reusable patterns and tooling to scale model optimizations to new hardware.
• Debug complex performance issues to deliver production-ready solutions.

Skills

PyTorch

Python

Transformer architectures

KVcache management

Distributed systems

Model optimization

Research leadership

Education

Bachelor's degree in CS/Engineering

Master's degree in CS/Engineering

PhD in CS/Engineering

Tools

TorchDynamo

Torch.compile

ONNX

Triton

Job description

Qualcomm Technologies, Inc. seeks a Staff Engineer to lead end-to-end AI model optimization for LLMs, VLMs, and diffusion on Qualcomm accelerators. You will transform PyTorch models, manage KVcache behavior, and drive deployment with PyTorch, ONNX, and torch.compile across multi-core systems.

You will collaborate with compiler and performance teams to craft lowering strategies and scalable tooling, ensuring high throughput with low latency and production-grade reliability.