Meta
Software Engineer, AI Kernels & Performance Optimization — MTIA Software
$184K–$257K
Sunnyvale, CA +3 · Full-time · Posted Aug 12, 2026
- Location
- Sunnyvale, CA
Bellevue, WA
Menlo Park, CA
New York, NY - Experience
- 6+ years in the listed fields; also states 8+ years in accelerator software, HPC, or ML systems performance, or equivalent with an advanced degree.
- Education
- Bachelor's degree in Computer Science, Computer Engineering, a related technical field, or equivalent practical experience.
About the job
Develops and optimizes kernels and authoring frameworks for Meta’s MTIA accelerators, from pre-silicon architecture analysis through production deployment.
What you'll do
- Design, optimize, and deploy compute and communication kernels for MTIA, profiling performance across the software and hardware stack.
- Build kernel frameworks and libraries that expand PyTorch operator coverage and enable other engineers to achieve high performance.
- Partner with silicon, compiler, runtime, and framework teams on hardware/software co-design, numerics, and production workload performance.
What they ask for
- Bachelor’s degree in Computer Science, Computer Engineering, a related technical field, or equivalent practical experience.
- 6+ years of professional experience in high-performance computing, accelerator kernel development, compiler backends, or systems performance engineering.
- Proficiency in C++ and Python, including low-level systems programming, templates, and generic programming.
- Demonstrated experience writing and optimizing kernels for a parallel architecture, plus working knowledge of computer architecture and measurement-driven performance analysis.
Nice to have
- Experience mentoring engineers and setting technical direction across teams.
- Experience with distributed execution and collective communication, low-precision numerics, compiler/codegen technologies, or pre-silicon development.
Pay
$183,997/year to $257,000/year USD base salary
Tools
C++, Python, PyTorch, CUDA, ROCm/HIP, SYCL/OpenCL, MLIR, LLVM
AI focus
Develops kernels and software frameworks for AI training and inference workloads, including recommendation, ranking, and generative AI; work includes attention kernels, low-precision numerics, and accelerator hardware/software co-design.
Model-written summary · the original posting is authoritative