Qualcomm
Embedded ML SW Architect -- Snapdragon Mobile (Principal or Director level)
$235K–$352K
San Diego, California · On-site · Principal · Full-time · Posted Oct 18, 2026
- Experience
- Bachelor's: 8+ years; master's: 7+ years; PhD: 6+ years. Also states 10+ years in High-Performance Computing System Engineering or Software, with 5+ years in AI system optimization.
- Education
- Bachelor's degree in Computer Science, Engineering, Information Systems, or related field; or master's or PhD in a related field. Preferred: master's or PhD in Computer Science or Electrical Engineering.
About the job
Architects and develops on-device AI software for Snapdragon Mobile, optimizing performance, power, memory, and heterogeneous compute across mobile hardware.
What you'll do
- Design AI features and on-device agentic experiences, including context, memory, perception, and CPU/GPU/NPU workload placement.
- Prototype solutions and optimize performance through profiling, benchmarking, telemetry-driven scheduling, and analysis of system constraints.
- Lead technical reviews and cross-functional alignment with internal teams, customers, OEMs, and external partners.
What they ask for
- Bachelor's degree in a related field and 8+ years of relevant experience, or master's and 7+ years, or PhD and 6+ years.
- 10+ years of experience in High-Performance Computing System Engineering or Software, with 5+ years in AI system optimization.
- Strong software engineering, performance analysis, and communication skills.
- Proficiency in C, C++, or Python.
Nice to have
- Experience with agentic AI, LLMs/SLMs, and embedded ML on resource-constrained devices.
- Experience with mobile platforms, AI frameworks such as TensorFlow or PyTorch, and heterogeneous CPU/GPU/NPU systems.
Pay
$234,800.00 - $352,200.00 USD base salary
Tools
Snapdragon Mobile, CPU, GPU, NPU, TensorFlow, PyTorch, C, C++
AI focus
Architects and optimizes embedded, on-device AI, including agentic experiences and LLM/SLM serving, workload scheduling, memory lifecycle, and AI performance and power efficiency.
Model-written summary · the original posting is authoritative