Senior Performance Architect, Nemotron

at Nvidia

USD 152,000-287,500 per year

SENIOR

✅ On-site

Tech Stack
Tag name is followed by "@" symbol and proficiency level value. About proficiency levels:

1-2 — basic awareness. Minimal hands-on experience, and a rudimentary understanding of the technology's purpose;

3-6 — daily use. Comfortable and regular usage, capable of handling common tasks and challenges related to the technology;

7-9 — you are an expert, you can teach others, you know all the pitfalls and tricks;

10 — exceptional knowledge, comprehensive understanding, and adeptness in all aspects of the technology, including advanced problem-solving. Think twice before claiming or demanding such level.

AI @ 4 Algorithms @ 6 CUDA @ 4 Data Analysis @ 6 Deep Learning @ 4 GPU @ 4 GenAI Generative AI @ 4 LLM @ 4 Machine Learning @ 4 Performance Analysis @ 7 PyTorch @ 4 Python @ 6 Reinforcement Learning @ 4 SGLang @ 4 vLLM @ 4

Details

NVIDIA is looking for a Senior Performance Architect for Nemotron to help shape the next generation of Nemotron models through performance modeling, analysis, and forward projections. The role focuses on AI model–system–hardware co-design and evaluating how architectural choices translate into real-world deployment efficiency. You will help ensure future models achieve Pareto-optimal trade-offs across accuracy, throughput, and interactivity on target platforms.

Recent efforts such as LatentMoE architectures and the Nemotron Super model exemplify the performance-driven co-design this role will advance. The position partners across research, framework development, compiler, and hardware teams to guide decisions affecting production-scale AI efficiency.

Responsibilities

Develop high-fidelity analytical performance models to prototype emerging algorithmic techniques and hardware optimizations for the Nemotron family of models.
Prioritize features and guide future software and hardware roadmaps based on detailed performance modeling and analysis.
Model the end-to-end performance impact of emerging generative AI workflows, including speculative decoding, agentic pipelines, inference-time compute scaling, and reinforcement learning, to understand future data center needs.
Keep up with the latest deep learning research and collaborate with deep learning researchers, hardware architects, and software engineers.

Requirements

A minimum of a master's degree, or equivalent experience, in computer science, electrical engineering, or a related field.
Strong background in computer architecture, roofline modeling, queuing theory, and statistical performance analysis techniques.
Solid understanding of machine learning fundamentals, model parallelism, and inference serving techniques.
Proficiency in Python; C++ experience is optional, for simulator design and data analysis.
At least 3 years of hands-on experience evaluating AI/ML workloads or performing performance analysis, modeling, and optimization for AI.
Experience defining metrics, designing experiments, and visualizing large performance datasets to identify resource bottlenecks.
Experience with deep learning frameworks such as PyTorch, TRT-LLM, vLLM, and SGLang.
A growth mindset and a pragmatic measure, iterate, and deliver approach.

Preferred Qualifications

A proven track record of working in multifunctional teams spanning algorithms, software, and hardware architecture.
Ability to distill complex analyses into clear recommendations for technical and non-technical collaborators.
Experience with GPU computing and CUDA.

Compensation and Benefits

Base salary for Level 3: USD 152,000–241,500 per year.
Base salary for Level 4: USD 184,000–287,500 per year.
Eligible employees may also receive equity and benefits.
NVIDIA is an equal opportunity employer committed to fostering a diverse work environment.