Inference Performance Engineer, Agent Driven Inference Optimization
Core
Senior Inference Performance Engineer optimizing large-scale AI inference workloads and serving frameworks to maximize GPU throughput and interactivity.
Role type
Senior IC Inference Performance Engineer
Builds
Optimized AI inference serving frameworks (TensorRT-LLM, SGLang, vLLM, Dynamo) and GPU kernels
Domain
AI Inference / GPU Computing / High-Performance Computing
Deliverable
production ML models
Required skills
GPU workload profiling, CUDA kernel optimization, Python engineering, C++/CUDA codebase navigation, experimental methodology, continuous batching, KV cache management, quantization, speculative decoding, MoE serving, NCCL/NIXL/NVSHMEM communication
Preferred skills
Agentic AI workflow automation, disaggregated serving, Hopper/Blackwell architecture expertise, MLPerf Inference benchmark results
Technologies
TensorRT-LLM, SGLang, vLLM, Dynamo, Nsight Systems, Nsight Compute, CUPTI, PyTorch profiler, CUDA, Hopper, Blackwell
Responsibilities
Design autonomous optimization workflows for AI agents; Profile and tune AI inference workloads for throughput and latency; Develop serving framework patches and optimized kernels; Collaborate with architecture and kernel teams to implement performance fixes.
Seniority
Senior, hands-on IC
