Senior GPU Performance Software Engineer

The Software and AI (SAI) organization is seeking a highly skilled software engineer to contribute to the development and low-level optimization of   oneDNN , a complex, cross-platform, open-source performance library that serves as the foundation for deep learning applications ( github.com/uxlfoundation/oneDNN ). Please Note:   This is a low-level software engineering and hardware-acceleration role. It does not involve building, training, or tuning machine learning models. Instead, you will focus on developing highly optimized math primitives, parallel algorithms, and GPU kernels that power industry-leading AI frameworks (such as OpenVINO, TensorFlow, PyTorch, and ONNX Runtime) on Intel hardware.

Kernel Development and Architecture Develop high-performance GEMM, convolution, and attention kernels for AI workloads Design scalable JIT and codegen infrastructure for GPU kernel generation Low-Level Optimization Implement fusion and memory-traffic optimizations to maximize hardware utilization Optimize mixed-precision and quantized execution paths (e.g., BF16, FP16, INT8, FP8, FP4, etc.) Performance Modeling and Profiling Build analytical and empirical performance models for kernel dispatch and tuning Profile and eliminate performance bottlenecks across oneDNN GPU primitives and runtime paths Hardware and Software Co-Design Co-design GPU primitives and kernel architectures for next-generation Intel GPUs Partner with hardware and compiler teams to shape future accelerator capabilities and software stacks Infrastructure and Validation Improve validation, benchmarking, and CI infrastructure for performance-critical GPU workloads Why Join Us Massive Scale Work on a global, high-impact open-source library that scales AI performance across millions of devices worldwide Cutting-Edge Hardware Get early access to and influence the software stack for Intel's roadmap of next-generation discrete GPUs Expert Collaboration Work alongside industry-leading experts in GPU compilers, hardware architecture, and performance libraries Total Rewards Enjoy a competitive package including stock programs, quarterly bonuses, robust healthcare, and highly flexible hybrid/remote working options What We're Looking For To be successful in this role, you should demonstrate the following professional traits: A strong ownership mindset — you take initiative on complex, ambiguous technical problems and drive them to resolution A collaborative approach — you work effectively across hardware, compiler, and framework teams to align on shared technical goals A performance-driven curiosity — you are motivated by squeezing every cycle out of hardware and continuously seek deeper understanding of low-level systems Qualifications:

Education:   BSc, MSc, or PhD in Computer Science, Computer Engineering, Mathematics, Physics, or a highly technical related field Core Language:   5+ years of professional software development experience with expert-level modern C++ Performance Optimizations:   2+ years of hands-on experience in programming and kernel optimization on GPUs (via SYCL/DPC++, OpenCL, CUDA, or HIP), or at least 5+ years of similar low-level performance optimization experience on CPUs Hardware Architecture:   Strong foundations in computer architecture, cache hierarchies, memory subsystems, and parallel programming paradigms (e.g., multi-threading, SIMD/vectorization) Preferred Qualifications Math Libraries:   Experience developing high-performance math libraries (e.g., GEMM, convolution, reduction, or FFT kernels) Low-Level Tuning:   Hands-on experience with GPU assembly-level tuning or compiler optimization Parallel APIs:   Familiarity with parallel programming APIs such as OpenMP or oneTBB AI Workload Context:   Basic understanding of deep learning primitives (e.g., forward/backward passes) to understand how library code is utilized