Remote ML Performance Engineer — Throughput Optimizer

Bright Vision Technologies; Renton (WA); On-site; USD 100,000 - 150,000; Full time

Job summary

Bright Vision Technologies is seeking an ML Performance Engineer to optimize large neural network workloads across training and inference. The role covers end-to-end performance from kernels to distributed systems, requiring deep GPU understanding and data-driven optimization.

You will work with product, design, and engineering teams to translate vague requirements into robust, scalable solutions. The ideal candidate has 6+ years in ML performance, strong Python/C++, and hands-on GPU

Qualifications

• 6+ years in performance engineering, ML systems, or HPC.
• Proficiency in Python and C++.
• Experience optimizing DL workloads on modern GPUs.
• Strong understanding of distributed training and inference.
• Experience with CPU/GPU profiling and distributed systems.
• Familiarity with model compression and accuracy implications.
• Strong memory hierarchy and parallelism knowledge.
• Excellent measurement, debugging, and analytical reasoning.
• Strong communication and collaboration skills.

Responsibilities

• Profile and optimize end-to-end AI training and inference for throughput, latency, and cost.
• Identify bottlenecks across data loading, compute, communication, and memory.
• Tune quantization, sparsity, pruning to reduce footprint and accelerate inference.
• Optimize distributed training with tensor/pipeline parallelism and sharding.
• Tune attention and memory optimizations for large models.
• Drive compiler-level optimizations using Triton, XLA, TorchInductor, TVM.
• Improve data pipelines, storage patterns, and throughput.
• Build and maintain benchmarks and regression tests.
• Collaborate with ML and platform teams to standardize pipelines.
• Advise on hardware selection and scheduling for cost efficiency.
• Document performance tuning findings and share with teams.
• Stay current with AI systems research and production translate.

Skills

Python

C++

GPU optimization

Profiling tools

Distributed training

ML systems

HPC

Measurement & debugging

Communication & collaboration

Education

Bachelors or Masters in CS/CE or related field

Tools

Triton

XLA

TorchInductor

TVM

CUTLASS

Job description

Bright Vision Technologies is seeking an ML Performance Engineer to optimize large neural network workloads across training and inference. The role covers end-to-end performance from kernels to distributed systems, requiring deep GPU understanding and data-driven optimization.

You will work with product, design, and engineering teams to translate vague requirements into robust, scalable solutions. The ideal candidate has 6+ years in ML performance, strong Python/C++, and hands-on GPU