Sr. Machine Learning Engineer

Prosum; Phoenix (AZ); Hybrid; USD 140,000 - 190,000; Full time

Job summary

Prosum seeks a Senior Machine Learning Engineer to architect and optimize real-time, high-throughput image pipelines for next-generation mask inspection tools. You will eliminate bottlenecks through CUDA kernel tuning and GPU parallel computing, ensuring deep learning models and CV algorithms process streaming data at production scale.

The role requires hands-on experience with TensorRT/ONNX, deployment in C++/Python applications, and collaboration with multiple engineering teams in a hybrid

Qualifications

• Senior ML/AI engineer with real-time processing experience.
• Experience architecting and optimizing image pipelines for high-throughput systems.
• Proven ability to deploy models in production-grade C++/Python applications.

Responsibilities

• Design and optimize real-time image pipelines for vision systems.
• Develop CUDA kernels and GPU-accelerated components.
• Deploy and optimize ML models with TensorRT/ONNX Runtime.
• Collaborate with ML scientists and software engineers.

Skills

CUDA

Python

C++

TensorRT

ONNX

GPU acceleration

Deep learning

Computer vision

Tools

GPU profiling

Real-time processing

Job description

Our client is seeking a Sr. Machine Learning Engineering for a direct hire role to sit in North Phoenix, AZ or Hillsboro, OR. This role will be onsite 4 days a week and 1 day remote.

JOB SUMMARY

The role of Senior Machine Learning Engineer will architect and optimize real-time, high-throughput, and ultra-low latency image pipelines for next-generation Mask Inspection Tools. Responsibilities include eliminating hardware bottlenecks through CUDA kernel tuning and GPU parallel computing, ensuring deep learning models and CV algorithms seamlessly processing massive, high-bandwidth streaming data at production scale.

ESSENTIAL DUTIES AND RESPONSIBILITIES

High-Performance Computing Pipeline Architecture

• Design, implement, and optimize high-throughput, low-latency image processing pipelines for real-time optical inspection and machine vision systems.
• Develop scalable architectures capable of processing large volumes of imaging data while meeting stringent latency and reliability requirements.
• Profile and optimize system performance across CPU, GPU, memory, and I/O subsystems.

GPU Acceleration

• Design, develop, and optimize CUDA kernels to accelerate deep learning inference and classical computer vision algorithms.
• Maximize GPU utilization through efficient memory management, kernel optimization, and parallel programming techniques.
• Evaluate and implement performance improvements using NVIDIA GPU technologies and profiling tools.

Model Deployment & Optimization

• Optimize, quantize, and deploy machine learning models using TensorRT, ONNX Runtime, or similar inference frameworks.
• Integrate AI models into production-grade C++ and Python applications.
• Improve inference throughput, latency, and resource utilization while maintaining model accuracy.
• Develop automated deployment and validation pipelines for machine learning models.

Concurrency & Systems Optimization

• Architect and implement multi-threaded, high-concurrency software components for data acquisition, buffering, streaming, and real-time processing.
• Design robust synchronization and communication mechanisms between hardware interfaces and AI processing pipelines.
• Optimize end-to-end system performance for deterministic, real-time execution.

Cross-Functional Collaboration

• Partner with machine learning scientists, computer vision engineers, hardware engineers, and software developers to deliver integrated AI solutions.