LLM Performance Engineer — GPU & HPC

Baseten; San Francisco (CA); On-site; USD 160,000 - 200,000; Full time

Benefits offered by this job

Competitive compensation

Equity

Medical/dental/vision insurance

Flexible PTO

Parental leave

401(k)

Learning opportunities

Job summary

Baseten is hiring early-career Software Engineers to join a team at the intersection of high-performance computing and large language model engineering. You will build automated speedometer and diagnostic tools for GPU clusters, measure FPS and memory bandwidth, and develop dev environments for model experimentation.

You will work across Linux, Python, and NVIDIA software to push performance and reliability in production-grade AI infrastructure.

Qualifications

• Interest in HPC and LLM engineering at scale.
• Willingness to learn and explore across hardware and software.
• Familiarity with Python and eagerness to master NVIDIA stack.

Responsibilities

• Run and automate LLM quality benchmarks and performance suites.
• Create automated tests for GPU clusters across x86 and ARM.
• Develop GPU-enabled dev environments optimized for model experimentation.
• Build and contribute to tools to automate model evaluation and optimization.
• Profile performance with PyTorch Profiler and Nsight Systems.
• Develop real-time dashboards and CI/CD for performance tests.
• Identify optimal configurations balancing latency, cost and quality.

Skills

Python

NVIDIA software stack

C++ (nice to have)

GPU architecture

Automation & scripting

Transformers / FLOPs concepts

Tools

PyTorch Profiler

NVIDIA Nsight Systems

genai-bench

InferenceMAX

CI/CD pipelines

Job description

Baseten is hiring early-career Software Engineers to join a team at the intersection of high-performance computing and large language model engineering. You will build automated speedometer and diagnostic tools for GPU clusters, measure FPS and memory bandwidth, and develop dev environments for model experimentation.

You will work across Linux, Python, and NVIDIA software to push performance and reliability in production-grade AI infrastructure.