Nvidia Senior Software Engineer - Automated Parallel Programming 

United States, Texas 

994155193

What you will be doing:

• Crafting a code generation system to accelerate portions of a graph collected from a machine learning framework.

• Partnering with NVIDIA’s hardware and software teams to improve GPU performance in PyTorch.

• Design, build and support production AI solutions used by enterprise customers and partners.

• Optimize the performance of influential, modern Deep Learning models coming out of academic and industry research, for NVIDIA GPUs and systems.

• Collaborating with internal applied researchers to improve their AI tools.

• Advise design of new hardware generations.

What we need to see:

• MS or PhD Computer Science, Computer Engineering, Electrical Engineering or a related field (or equivalent experience).

• Parallel programming experience with writing optimized kernels in the NVIDIA CUDA Programming Language or similar parallel languages

• 4+ years of experience with C++ programming.

• Demonstrated experience developing large software projects.

• We require excellent verbal and written communication skills.

Ways to stand out from the crowd:

• Proven technical foundation in CPU and GPU architectures, numeric libraries, modular software design.

• A background in deep learning compilers or compiler infrastructure

• Expertise with optimized distributed parallelism techniques and it's a bonus if that includes parallelizing Large Language Models!

• Knowledge of heuristic generation that employs cost models, machine learning, or auto-tuning.

• Contributions to PyTorch, Numpy, JAX, TensorFlow, OpenAI-Triton, Lightning Thunder, TVM, Halide or similar system.