Systems ML Engineer: AI Compiler & Kernel Dev

Meta; Seattle (WA); On-site; USD 122,000 - 181,000; Full time

Job summary

Meta’s MTIA Software Team is hiring a Software Engineer to work on the PyTorch AI framework core compilers and AI accelerator kernels. You will analyze deep learning models, implement optimizations, and collaborate with AI researchers to accelerate next‑gen models across recommendations, vision and NLP.

The role focuses on high‑performance software and hardware co-design. You will gain experience in compiler frameworks, DL graph optimizations, and kernel authoring for MTIA hardware, contributing

Qualifications

• Bachelor's degree in CS/CE or equivalent, completed before joining Meta.
• Proven C/C++ programming skills.
• Experience in AI framework development or accelerating DL models on hardware.
• Familiarity with PyTorch, Caffe2, TensorFlow, ONNX or TensorRT is a plus.

Responsibilities

• Develop the software stack for AI compiler, frameworks and high‑performance kernels.
• Contribute to PyTorch AI framework core compilers for new hardware accelerators.
• Analyze deep learning models and implement compiler optimization algorithms.
• Collaborate with AI researchers to accelerate next‑gen DL models across domains.
• Perf tuning and optimization of DL framework components.

Skills

C/C++ programming

AI framework development

Hardware acceleration experience

Education

Bachelor's degree in Computer Science or Computer Engineering or equivalent

Tools

PyTorch

TensorFlow

ONNX

Caffe2

TensorRT

CUDA

Job description

Meta’s MTIA Software Team is hiring a Software Engineer to work on the PyTorch AI framework core compilers and AI accelerator kernels. You will analyze deep learning models, implement optimizations, and collaborate with AI researchers to accelerate next‑gen models across recommendations, vision and NLP.

The role focuses on high‑performance software and hardware co-design. You will gain experience in compiler frameworks, DL graph optimizations, and kernel authoring for MTIA hardware, contributing