Apple
ML Framework (MetalLM) Engineer, Graphics, Game and ML
$150K–$278K
Cupertino, California · Full-time · Posted Mar 27, 2026
- Experience
- 3+ years of programming and problem-solving experience.
About the job
Build and optimize distributed machine-learning inference frameworks for high-throughput GPU execution on Apple server hardware.
What you'll do
- Optimize scalable ML inference using distributed compute strategies, including data, tensor, pipeline, and expert parallelism.
- Develop kernel- and compiler-level optimizations and analyze performance across server hardware families.
- Apply model optimization techniques and improve latency, memory footprint, and compute efficiency.
What they ask for
- 3+ years of programming and problem-solving experience with C/C++/ObjC.
- Experience developing and optimizing GPU kernels using compute programming models such as Metal or CUDA.
- Experience with distributed training or inference techniques.
- Experience with system-level programming and computer architecture.
Nice to have
- Experience with graph compilers such as CuTE, CuTile, Triton, OpenXLA, or LLVM.
- Good understanding of LLM and diffusion-based model architectures.
Pay
$150,400 and $277,600 (USD, base pay)
Tools
C/C++/ObjC, Metal, CUDA, CuTE, CuTile, Triton, OpenXLA, LLVM
AI focus
Develop and optimize distributed GenAI inference, including model optimization techniques such as speculation, quantization, and compression.
Model-written summary · the original posting is authoritative