Senior Software Engineer, CUTLASS Performance

Description

The company builds high-performance computing platforms that power AI across many applications and industries.

CUTLASS is an open-source library, part of the company's software stack, for high-performance linear algebra and Tensor Core primitives. Since 2017 it has given the community C++ and Python abstractions for writing custom matrix multiply (GEMM) and related math and deep learning computations on the company GPUs.

This role suits someone who cares deeply about performance and wants to close the gap between current results and what is theoretically achievable on the CUTLASS team.

You would join the Deep Learning Library team, which builds a real-time, cost-effective computing platform for a fast-growing field. The company is widely regarded as one of the most desirable technology employers and its staff are known for being forward-thinking and hard working. The team wants people who are creative, self-directed and up for a challenge.

Responsibilities

• Measure the performance of leading deep learning models during inference and training to find GPU kernel and fusion opportunities.
• Locate where realized performance falls short of theoretical limits, then propose software changes or model adjustments to close the gap.
• Build tools that automate benchmarking, analysis and optimization, so CUTLASS kernels perform as well as possible inside deep learning networks.
• Serve as the team's go-to expert on kernel performance and act as the CUTLASS performance contact for GPU architecture, deep learning framework, and QA teams across the company.

Requirements

• A master's or PhD in computer science, computer engineering or a related field, or equivalent experience.
• At least 3 years of relevant industry experience.
• Strong Python and C++ programming skills.
• Experience analysing and optimising software performance.
• Solid grasp of computer architecture and working knowledge of GPUs or other parallel processing architectures.
• Nice to have: deep knowledge of current deep learning model architectures.
• Nice to have: hands-on experience benchmarking deep learning frameworks such as PyTorch, JAX, SGLang, vLLM, TRT-LLM or others.
• Nice to have: experience building performance models and performance regression systems.

Conditions

Base salary depends on location, experience and other objective job-related factors. Level 3 ranges from 152,000 USD to 241,500 USD; Level 4 ranges from 184,000 USD to 287,500 USD.

Applications are accepted at least until November 3, 2026.

This is an existing vacancy.

The company uses AI tools in parts of its recruiting process.

The company is committed to an inclusive workplace and does not discriminate in hiring or promotion on any basis protected by law.