Architect, GPU Performance at NVIDIA (Bengaluru)

September 12, 2026
Application ends: October 12, 2026
Apply Now

Job Description

NVIDIA is hiring an Architect for GPU performance in Bengaluru to analyse and improve the system-level performance of its GPUs and systems-on-chip, using performance models, RTL test benches, emulators and silicon to find bottlenecks and inform architecture decisions.

At a glance

  • Role: Architect – GPU Performance
  • Location: Bengaluru
  • Experience: 3+ years in performance analysis of SoC or GPU architectures
  • Focus: system-level performance analysis, workloads and performance modelling
  • Requisition: JR2017247
  • Apply: open until filled (re-checked 12 October 2026)

What you will do

  • Analyse system-level performance and bottlenecks of complex GPUs and SoCs
  • Use performance models, RTL test benches, emulators and silicon to find and explain bottlenecks
  • Develop workloads and test suites for graphics, machine learning, automotive, video and computer-vision use cases
  • Explore architecture trade-offs in performance, area and power with architecture and design teams
  • Build performance models, testbench components, and analysis and visualisation tools
  • Drive methods that shorten turnaround time and enable earlier performance analysis

Performance architecture sits between chip research and product design: engineers build models of hardware that does not exist yet, run realistic graphics, machine-learning, automotive, video and vision workloads on them, and feed the results back into trade-offs between speed, area and power. It suits people who enjoy computer architecture, simulation and data analysis.

What NVIDIA is looking for

  • A BE/BTech or MS/MTech in a relevant area, or equivalent experience (a PhD is a plus)
  • 3+ years with performance analysis and complex SoC or GPU architectures
  • Strong understanding of SoC architecture, the graphics pipeline, memory subsystems and network-on-chip interconnects
  • Expert C/C++ and scripting in Perl or Python (Verilog/SystemVerilog or SystemC/TLM a strong plus)
  • Strong debugging and statistical analysis skills, including using RTL dumps
  • Experience building performance simulators or cycle-accurate/approximate models (a strong plus)

Pay and location

NVIDIA’s posting does not state a salary. The role is based in Bengaluru.

How to apply

Apply through the official NVIDIA job posting. NVIDIA’s posting does not give a closing date, so the role is open until filled; ResearchJobs.in will re-check this listing on 12 October 2026. Check the posting for location and work-authorisation details before you apply.

See all our engineer jobs, or browse more research jobs on ResearchJobs.in.

Hiring institution: NVIDIA

Official advertisement: nvidia.wd5.myworkdayjobs.com

How to prepare for this application

  • Architecture fundamentals: revise caches, memory controllers, NoC routing and the GPU pipeline.
  • Modelling: be ready to explain how you would build a cycle-approximate model of a subsystem.
  • Workloads: think about which benchmarks stress memory bandwidth versus compute.
  • Data analysis: prepare an example of finding a bottleneck from simulation traces.
  • Trade-offs: practise reasoning about performance, area and power together.

About NVIDIA

NVIDIA, headquartered in Santa Clara, California, designs graphics processing units (GPUs) and accelerated computing platforms used for AI, graphics, scientific computing, autonomous vehicles and robotics. Its DRIVE, Jetson and IGX platforms, mentioned in several of its India roles, target cars, robots and industrial edge systems. NVIDIA has large engineering teams in India, with offices in Bengaluru, Pune, Hyderabad and Gurugram.

We send one confirmation email first. Every alert has an unsubscribe link.