This role pays more than
52%
of similar roles. Most pay
$162,000–$235,750
— the shaded band above.
At the midpoint, this role pays about
$198k
versus about
$199k
for comparable roles.
Based on 240 similar postings.
Employer
About Qualcomm
Qualcomm is a leading American semiconductor and telecommunications company based in San Diego, CA.
Qualcomm currently has
623 open roles
on FindRole.
Listed pay typically runs
$148,300–$222,500
across 603 roles with salary data.
As an AI Model Optimization Architect on the Machine Learning Engineering team, you will lead end-to-end model transformation and optimization for LLMs, VLMs, diffusion, and multimodal models on inference accelerators. You will be responsible for developing optimization strategies to transform PyTorch models into efficient executions while balancing throughput, latency, memory, and quality. Your daily work involves driving graph capture using PyTorch, ONNX, and torch.compile, designing fusion kernels via Triton, and managing KVcache and continuous batching systems. You will collaborate with compiler and performance teams to solve complex technical challenges regarding distributed inference and hardware-specific scaling. The role requires expert proficiency in PyTorch, Python, and transformer architectures, alongside a deep understanding of computer architecture, ML accelerators, and distributed systems to ensure production-ready solutions for large-scale foundation models across various hardware platforms.
Architect and deliver model optimization strategies to transform PyTorch models for efficient inference on Qualcomm accelerators.
Drive graph capture and deployment using PyTorch, ONNX, and torch.compile through model rewrites and transformations.
Design and implement fusion kernels using DSL-based approaches like Triton for performance-critical algorithmic rewrites.
Profile and optimize LLM, VLM, and diffusion inference for throughput and latency across various serving modes.
Manage transformer-specific optimizations including KVcache management, decoding behavior, and long context performance.
Enable and optimize continuous batching systems to manage memory, scheduling, and tail latency.
Architect and scale distributed inference strategies such as sharding and parallelism across multi-core and multi-device systems.
Establish reusable optimization patterns and tooling to scale model optimizations across new hardware architectures.
What we're looking for
Expert level expertise in PyTorch and inference focused model optimization with strong Python engineering skills.
Hands-on experience with torch.compile, TorchDynamo, or related graph capture and compilation workflows.
Deep understanding of transformer architectures, attention mechanisms, MoEs, and performance trade-offs.
Practical experience with KVcache behavior, serving time optimizations, and memory/performance tradeoffs.
Strong foundation in computer architecture, ML accelerators, and distributed systems.
Proven ability to lead cross-functional technical efforts and influence design decisions.
Bachelor's degree and 4+ years of relevant engineering experience, or a Master's degree and 3+ years of experience, or a PhD and 2+ years of experience.
LLM
PyTorch
Python
Triton-Inference Server
vLLM
SGLang
CUDA
Triton
torch.compile
torchDynamo
Distributed Systems
Kernel Design
Deep Learning
KV-Cache Management
Model Optimization