Senior Embedded ML Systems Architect (Mobile AI)

QUALCOMM, Inc.; San Diego (CA); On-site; USD 235,000 - 352,000; Full time

Job summary

Qualcomm Technologies, Inc. seeks an experienced Embedded ML SW Architect focused on Snapdragon Mobile products to drive AI software architecture and system design for high-performance, power-efficient mobile AI solutions.

You will influence the SW stack from CPU/GPU/NPU to memory lifecycle, work with cross-functional teams, and push innovations in agentic AI systems and edge AI deployments.

Qualifications

• Bachelor's degree with 8+ years of Hardware/Software/Systems Engineering or related field.
• Master's degree with 7+ years of Hardware/Software/Systems Engineering or related field.
• PhD with 6+ years of Hardware/Software/Systems Engineering or related field.

Responsibilities

• Develop architecture for AI features on Snapdragon Mobile products; recommend architecture enhancements.
• Independently research and work cross-functionally to define requirements. Hands on prototype development.
• Architect on-device agentic experiences, including memory management and heterogeneous compute placement.
• Architect memory model-state lifecycle management for on-device LLM/SLM serving.
• Evaluate memory bandwidth and data-movement constraints to maximize AI throughput.

Skills

Embedded ML

Performance optimization

CPU/GPU/NPU orchestration

Android/mobile platforms

C/C++ programming

Python

System design

Leadership

Strong communication

Problem solving

Education

Bachelor's degree

Master's degree

PhD

Tools

TensorFlow

PyTorch

LoRA/adapter-based fine-tuning

LLMs/SLMs

Job description

Qualcomm Technologies, Inc. seeks an experienced Embedded ML SW Architect focused on Snapdragon Mobile products to drive AI software architecture and system design for high-performance, power-efficient mobile AI solutions.

You will influence the SW stack from CPU/GPU/NPU to memory lifecycle, work with cross-functional teams, and push innovations in agentic AI systems and edge AI deployments.