AI Inference Optimization Engineer

What you'll do

• Deploy, profile, and optimize vision models, LLMs, VLMs, and VLAs on GPNPU, including kernel and compute-graph tuning.
• Evaluate accuracy and select quantization strategies; integrate pruning, sparsification, and quantization into deployment workflows.
• Build profiling, tracing, visualization, and automation tools to identify bottlenecks and streamline model deployment.

What they ask for

• 3+ years of experience in AI model deployment or kernel optimization.
• BS/MS in CS/CE or related fields.
• Strong C/C++ and Python; hands-on performance analysis using profilers and compute-versus-memory bottleneck reasoning.
• Hands-on quantization experience (PTQ/QAT, int8/int4), debugging accuracy loss, and writing performance-critical kernels using CUDA, Triton, OpenCL, SIMD intrinsics, or DSP; fluent with AI coding agents.

Nice to have

• Hands-on experience in pruning and sparsification.
• Understanding of accelerator microarchitecture, fixed-point and numerics, mixed precision, or compiler code generation and kernel scheduling.

Pay

USD 160,000 to USD 195,000 base salary

Tools

GPNPU, C/C++, Python, CUDA, Triton, OpenCL, SIMD intrinsics, DSP

AI focus

Run and optimize AI model inference on an edge GPNPU, including CNN perception models, LLMs, VLMs, and VLAs; use quantization, pruning, sparsification, and agent-driven deployment workflows.