As a Staff Engineer – AI Accuracy Architect at Qualcomm Technologies, you will lead the development of accuracy-centric architecture and optimization for large language models (LLMs), vision-language models (VLMs), and multimodal models, collaborating closely with compiler, performance, and model optimization teams. Your daily tasks include designing quantization strategies such as PTQ, QAT, mixed precision, and per-channel/group-wise approaches to ensure optimal accuracy while balancing performance and hardware constraints. You will analyze numerical stability issues across various layers of the inference stack, from kernels to runtimes, and work with PyTorch and ONNX for model conversion and deployment. This role requires expertise in transformer architectures, attention mechanisms, and mixed-precision training, along with strong Python skills and a solid understanding of computer architecture and ML accelerators.
Us, Ca, Santa Clara, US55 days ago
$152,000–$241,500
Python
TypeScript
Go
Rust
C++
Synthetic data generation
GPU-backed inference systems
Multi-agent workflows
Orchestration patterns
Complex long-running task systems
MCP
A2A-style communication patterns
Secure execution
Sandboxing
Secrets handling
Auditability
CI/CD