Senior Cosmos Infra Engineer - End-to-End Performance

NVIDIA AI; Santa Clara (CA); On-site; USD 180,000 - 260,000; Full time

Benefits offered by this job

Equity

Benefits

Job summary

NVIDIA AI seeks a highly skilled professional to build state-of-the-art World Foundation Models and drive end-to-end performance analysis for data center and edge deployments. You will develop infrastructure to automate data ingestion, curation, pre-training, and deployment for Physical AI systems.

The role requires a Master’s degree in a STEM field and 5+ years of experience with large-scale parallel distributed systems and AI workload optimization, with proficiency in Python, C++, and

Qualifications

• Requires a Master's degree in a STEM field and 5 years of experience with large-scale parallel distributed systems and AI workload optimization.
• Proficiency in Python, C++, and Distributed PyTorch is essential, along with a deep understanding of World Foundation Models.

Responsibilities

• Build state-of-the-art World Foundation Models and drive end-to-end performance analysis for data center and edge deployments.
• Develop infrastructure to automate data ingestion, curation, pre-training, and deployment for Physical AI systems.

Skills

Distributed PyTorch

Python

C/C++

CUDA

Performance Modeling

Computer Architecture

Distributed Systems

World Foundation Models

Quantization

Data Curation

Transformer Models

Cloud Service Providers

DNNs

Networking

Storage Systems

Edge Deployment

Education

Master's degree in STEM

Job description

NVIDIA AI seeks a highly skilled professional to build state-of-the-art World Foundation Models and drive end-to-end performance analysis for data center and edge deployments. You will develop infrastructure to automate data ingestion, curation, pre-training, and deployment for Physical AI systems.

The role requires a Master’s degree in a STEM field and 5+ years of experience with large-scale parallel distributed systems and AI workload optimization, with proficiency in Python, C++, and