GM is seeking to build and scale robust Compute platforms for ML workflows to support autonomous vehicles and AI-driven products, and to optimize for high-priority, ML-centric use cases.
Requirements
- Expertise in either Go, Python, C++ or other relevant coding languages.
- Expertise in ML inference, model serving frameworks (triton, rayserve, vLLM etc).
- Experience working with cloud platforms such as GCP, Azure, or AWS.
- Hands-on experience building ML infrastructure platforms for model serving/inference.
- Experience working with or designing interfaces, apis and clients for ML workflows.
- Experience with Ray framework, and/or vLLM.
- Familiarity with hardware acceleration (GPUs) and optimizations for inference workloads.
Responsibilities
- Design and implement core platform backend software components.
- Collaborate with ML engineers and researchers to understand critical workflows, parse them to platform requirements, and deliver incremental value.
- Lead technical decision-making on model serving strategies, orchestration, caching, model versioning, and auto-scaling mechanisms.
- Drive the development of monitoring, observability, and metrics to ensure reliability, performance, and resource optimization of inference services.
- Proactively research and integrate state-of-the-art model serving frameworks, hardware accelerators, and distributed computing techniques.
- Lead large-scale technical initiatives across GM’s ML ecosystem.
- Raise the engineering bar through technical leadership, establishing best practices.
Other
- 8+ years of industry experience, with focus on machine learning systems or high performance backend services.
- Strong communication skills and a proven ability to drive cross-functional initiatives.
- Ability to thrive in a dynamic, multi-tasking environment with ever-evolving priorities.
- Bachelor's degree or higher in a relevant field (not explicitly mentioned but implied)
- Travel to the GM Global Technical Center - Cole Engineering Center Podium, MI or Mountain View Technical Center, CA at least three times per week
