Server Performance Architect - Hardware

Role overview

Server Performance Architect - Hardware at NVIDIA

Defining and driving server-level performance targets across CPU, GPU, memory, interconnect, networking, and storage subsystems

This position demands a unique capability to bridge deep architectural knowledge, workload analysis, and hands-on silicon investigations.

The role does not heavily involve modeling tasks (functional or performance), though occasional focused assignments may arise.

Responsibilities

• Conducting hands-on workload characterisation and bottleneck analysis on NVIDIA and competitive server platforms using AI training, inference, and HPC benchmarks
• Leveraging profiling, tracing, and analysis tools to root-cause performance issues and identify optimisation opportunities at the system level
• Performing trade-off studies on system topology, thermal/power envelopes, and memory hierarchy to guide architectural decisions
• Collaborating with silicon, platform, firmware, and software teams to identify and close performance gaps from bring-up through production
• Developing automation and tooling for performance regression tracking and reporting
• Representing the performance perspective in architecture reviews and cross-functional design discussions
• Building and maintaining analytical performance models and simulation frameworks for next-generation server platforms
• Publishing internal performance studies and best-practice guides for partner and customer enablement
• What we need to see:
• BS, MS, or PhD in Electrical/Computer Engineering, Computer Science, or equivalent experience
• 10+ years of experience in server/system performance architecture or related disciplines
• Deep understanding of modern server architectures — CPU microarchitecture, PCIe/CXL, DDR/HBM memory subsystems, and coherency protocols
• Strong hands-on experience with system-level profiling and performance analysis tools on server platform(s)
• Solid knowledge of GPU-accelerated compute, high-performance networking, or high-performance storage subsystems
• Proficiency in Python, C/C++, or similar languages for scripting, data analysis, and tool development
• Comfortable and proficient using AI-powered coding and productivity tools to accelerate analysis, automation, and documentation workflows
• Excellent communication skills with the ability to distil complex performance data into actionable architectural recommendations
• Ways to stand out from the crowd:
• Experience with AI/ML training and inference workloads at data-centre scale
• Familiarity with NVIDIA GPU architectures (Hopper, Blackwell, Rubin) and associated software stacks (CUDA, NCCL)
• Background in chip-to-chip interconnect performance analysis (C2C, UCIe)
• Exposure to power/thermal-aware performance optimisation techniques
• Track record of contributions to industry conferences or published performance studies
• #LI-Hybrid

Employment details

• Location: India [Bengaluru], India [Gurugram], India [Hyderabad], India [Pune]

About the employer

• NVIDIA has continuously reinvented itself.
• NVIDIA is a “learning machine” that constantly evolves by adapting to new opportunities that are hard to solve, that only we can address, and that matter to the world.
• NVIDIA is seeking architects to drive architectural performance for its next-generation AI server systems.

Additional role details

• Our invention of the GPU sparked the growth of the PC gaming market, redefined modern computer graphics, and revolutionized parallel computing.
• Today, research in artificial intelligence is booming worldwide, which calls for highly scalable and massively parallel computation horsepower that NVIDIA GPUs excel.
• This is our life’s work , to amplify human creativity and intelligence.
• As an NVIDIAN, you’ll be immersed in a diverse, supportive environment where everyone is inspired to do their best work.
• Come join our diverse team and see how you can make a lasting impact on the world!
• Candidates should be adept at working directly with silicon, high-level models, and simulators.
• Responsibilities include conducting performance investigations on both NVIDIA and competitive platforms, and developing targeted microbenchmarks to examine specific architectural aspects.