Senior AI Inference Architect for Scalable Services

NVIDIA Corporation (old); Santa Clara (CA); Hybride; USD 184 000 - 357 000; Plein temps

Résumé du poste

NVIDIA is seeking an architect to redefine AI inference across large-scale GPU deployments. You will lead the design of reference architectures, guide partners, and drive performance gains in disaggregated inference, with emphasis on scalable, codified best practices.

You will mentor teams and collaborate with customers to implement end-to-end inference solutions, leveraging Dynamo, Triton, TensorRT-LLM, and Kubernetes-driven workflows.

Qualifications

• 6+ years in solutions architecture driving customer engagements to deploy distributed systems.
• 2+ years with AI workloads on Kubernetes.
• Experience with Dynamo, Triton or TensorRT-LLM for model optimization and serving.
• Hands-on full-stack agent design with memory and retrieval systems.
• BS in CS/Engineering or equivalent experience.

Responsabilités

• Work with inference partners to build value and share product insights.
• Build and operate inference recipes using distributed GPU workers to improve efficiency.
• Accelerate pipelines with TensorRT-LLM and related backends for seamless integration.
• Advise on devops practices, Kubernetes clusters, compute fabrics, and observability.
• Provide mentorship and technical leadership for deployments of disaggregated inference systems.

Connaissances

Solutions architecture

AI workloads on Kubernetes

Inference optimization

Full-stack design

Leadership & mentoring

Formation

Bachelor's degree in CS/Engineering

Outils

NVIDIA Dynamo

Triton Inference Server

TensorRT-LLM

Description du poste

NVIDIA is seeking an architect to redefine AI inference across large-scale GPU deployments. You will lead the design of reference architectures, guide partners, and drive performance gains in disaggregated inference, with emphasis on scalable, codified best practices.

You will mentor teams and collaborate with customers to implement end-to-end inference solutions, leveraging Dynamo, Triton, TensorRT-LLM, and Kubernetes-driven workflows.