Senior Machine Learning Engineer - ML Training Infrastructure

General Motors is seeking an experienced Senior Machine Learning Engineer to design and build scalable AI/ML platform infrastructure. The role involves collaborating with teams to develop advanced AI solutions for intelligent driving technologies.

Responsibilities:

• Design and development of scalable, reliable, high-performance ML framework to support model training at scale
• Model training performance analysis and optimization solutions to scale distributed training workflows and maximize resource utilization across heterogeneous hardware environments, and save cost
• Raise the bar on system observability, debuggability, and operational excellence, and user experience
• Collaborate with cross-functional teams to integrate new features and technologies into the platform

Requirements:

• Bachelors degree or higher in Computer Science or equivalent major OR equivalent relevant experience
• 2+ years professional software engineering experience
• 2+ years specialized experience in AI/ML infrastructure, e.g., enabling distributed training for scaling large ML models
• Strong programming skills in Python, with proficiency in frameworks such as, PyTorch (preferred), TensorFlow, or similar
• Experience with distributed computing, GPU computing, and cloud environments (AWS, GCP, Azure)
• Willingness to travel to Sunnyvale, CA as needed
• Comfortable working in highly ambiguous and dynamic environments
• 3+ years of professional software engineering experience
• Self-motivated, strong execution, impact-delivering oriented
• Extensive knowledge and experience with PyTorch 2.x+ and distributed training framework
• Experience with design and development of training framework that supports FSDP, Pipeline Parallelism and other scalable solutions to training large foundational models
• Experience with profiling, analysis, debugging and optimizing training and data loading performance
• Excellent communication skills to resolve controversial, make consensus, communicate risks and give constructive feedback