AI Infrastructure Engineer

Intel•Hillsboro, CA

5d•Hybrid

About The Position

We are looking for a performance-obsessed AI Infrastructure Engineer to push LLM inference to its absolute limits on Intel's next-generation GPU architectures. In this role, you will dive deep into the inference stack and redefine peak performance. You will work end-to-end across the stack: profiling bottlenecks, writing custom GPU kernels, and upstreaming your optimizations directly into industry-standard serving frameworks like vLLM and SGLang. Your optimizations will be instrumental in unlocking the full potential of Intel hardware for state-of-the-art generative AI workloads.

Requirements

Bachelors Degree in Computer Science, Software Engineering, Artificial Intelligence/Machine Learning, or related field and 4+ years experience, Masters Degree and 3+ years, OR PhD.
3+ years of relevant software engineering experience in GPU computing, AI systems, or high-performance computing (HPC).
Proficiency in modern C++ and Python. You are comfortable reading and modifying complex systems-level code.

Nice To Haves

Understanding of CPU/GPU architecture.
Understanding of modern LLM architectures and inference paradigms: attention mechanisms, KV caching, continuous batching, speculative decoding, and prefill-decode disaggregation.
Prior open-source contributions to inference engines (vLLM, SGLang, PyTorch, llama.cpp).
Hands-on experience writing and optimizing custom GPU kernels using Triton, SYCL, CUDA/CUTLASS, or other DSLs.
Experience with scale-out inference orchestration across multi-node topologies.
You leverage AI coding agents daily to accelerate your own workflow and benchmark generation.

Responsibilities

Drive Inference Performance: Own the end-to-end optimization pipeline for running state-of-the-art LLMs on Intel GPUs.
Deep Stack Optimization: Profile, diagnose, and resolve cross-stack performance bottlenecks.
Kernel Development and Integration: Design, write, and optimize custom high-performance kernels for critical attention mechanisms, MoE, quantization, and operator fusions.
Open Source Leadership: Upstream your architectural improvements and hardware backends directly into open-source repositories like vLLM, SGLang, and PyTorch, acting as a bridge between the hardware teams and the open-source community.
Shape the Hardware Roadmap: Apply roofline analysis and systematic profiling to decompose bottlenecks. You will partner with our architecture and compiler teams to shape future GPU roadmaps based on real-world GenAI workload data.
Show passion about AI infrastructure and performance optimization.