Senior ML Kernel Optimization Engineer

Amazon; Toronto; Hybrid; CAD 151,000 - 252,000; Full time

Benefits offered by this job

Health insurance

RRSP

DPSP

Paid time off

Job summary

Amazon's Annapurna Labs team is hiring engineers to design and optimize high-performance compute kernels for ML on Neuron. You will work across Neuron architecture generations, profiling and tuning for maximum throughput, and implement compiler optimizations while collaborating with customers to enable their models on AWS accelerators.

The role blends machine learning, high-performance computing, and distributed architectures in a startup-like environment, with mentorship and customer-facing

Qualifications

• 5+ years of full software development life cycle experience.
• Bachelor's degree in computer science or equivalent.
• Expertise in accelerator architectures for ML/HPC (GPUs/FPGA/ASIC).
• Experience with CUDA, OpenCL, ROCm and GPU backends.
• Proficiency with NVIDIA PTX and GPU ISA concepts.
• Experience developing high-performance HPC libraries.
• Strong low-level optimization for GPUs and ML frameworks.

Responsibilities

• Design and implement high-performance compute kernels for ML on Neuron.
• Optimize kernel performance across multiple Neuron generations.
• Perform detailed profiling to identify bottlenecks and improve throughput.
• Apply compiler optimizations: fusion, tiling, scheduling.
• Collaborate with customers to optimize models on AWS accelerators.
• Work cross-functionally to develop innovative kernel techniques.

Skills

SDLC experience

Accelerator architectures

GPU kernel optimization

NVIDIA PTX

High-performance libraries

GPU memory optimization

LLVM/MLIR backends

PyTorch / TensorFlow GPUs

Parallel programming

GPU backends knowledge

Education

Bachelor's degree

Tools

CUDA

OpenCL

ROCm

PyTorch

TensorFlow

LLVM

Job description

Amazon's Annapurna Labs team is hiring engineers to design and optimize high-performance compute kernels for ML on Neuron. You will work across Neuron architecture generations, profiling and tuning for maximum throughput, and implement compiler optimizations while collaborating with customers to enable their models on AWS accelerators.

The role blends machine learning, high-performance computing, and distributed architectures in a startup-like environment, with mentorship and customer-facing