Nvidia Software Development Engineer Test - Deep Learning 

United States, Texas 

457474053

What you’ll be doing:

• In this role you will work closely with compiler developers to test new and state of the art deep learning related features and components including crafting and executing unit, functional, and performance tests. This would include authoring and reviewing test plans, implementing & automating test cases and porting 3rd party tests.

• Generating test reports, isolating and classifying failures and tracking performance trends.

• Automate compiler testing using NVIDIA test frameworks and by programming. Includes test execution, test reporting, and results analysis and automation of build and test environments. Work with software compiler developers and assist in providing automated solutions for unit testing.

• Help identify potential or observed weaknesses in the current process, offer ideas for actions that can improve code coverage, and participate in quality initiatives and drive continuous improvement.

What we need to see:

• Bachelor’s or Master’s Degree or equivalent experience.

• 2+ years of strong programming skills in C++ and Python

• Knowledge of compiler design principles, optimization techniques, code generation, and familiarity with modern compiler architectures and toolchains

• Deep Learning Domain Knowledge: Solid understanding of deep learning concepts, particularly with transformers and large language models (LLMs).

• Experience with writing test plans, test development, test automation, test execution and reporting in a production environment.

• You should be focused, learn quickly, and have strong analytical skills with attention to detail. Strong troubleshooting and debugging skills.

• Demonstrated uses of creative thinking for solutions to exciting problems that matter.

• Excellent communications skills, self-motivated and well organized.

Ways to stand out from the crowd:

• Experience with machine learning domain-specific languages (DSLs), including tile-based languages like Triton for efficient GPU programming and optimization.

• Familiarity with the fundamentals of the LLVM compiler infrastructure and MLIR (Multi-Level Intermediate Representation), with an understanding of how they function and their applications in optimizing and developing custom compiler tools.

• Familiarity with CUDA for parallel programming on NVIDIA GPUs, with a solid understanding of GPU architecture, memory management, and optimization techniques for enhancing performance in computational tasks.

• Previous compiler development and/or compiler verification/test or performance analysis experience.

• Experience with NVIDIA CUDA Toolkit, especially solving issues and debugging in the Linux environment.