Staff AI Inference Engineer — Production Optimizations

Crusoe; United States; On-site; USD 215,000 - 260,000; Full time

Benefits offered by this job

Competitive compensation

Restricted Stock Units

Paid time off, paid holidays & leave

Health insurance (comprehensive)

HSA contributions

Paid parental leave

Life insurance and disability

Professional development & tuition

Mental health & wellness support

Commuter benefits

Cell phone stipend

401(k) Retirement plan with company 4%

Volunteer time off

Travel insurance & emergency support

Daily meals allowance

Location-specific perks

Job summary

Crusoe is seeking a hands-on engineer to make large language models run faster, cheaper, and more reliably in production. You will own the inference stack end to end, from profiling time and cost to optimizing CUDA kernels and deployment strategies.

You will work with customer teams to tailor deployments, ship production-ready improvements, and drive performance across diverse models using Python as a primary language.

Qualifications

• Bachelor's, Master's, or Ph.D. in CS/Engineering/Math or related field.
• Hands-on production code experience in Python or C++ (prefer Python).
• Experience optimizing LLMs for high throughput / low latency.
• Familiar with LLM serving frameworks such as vLLM or SGLang; profiling to kernel level.
• Strong understanding of GPU architecture and behavior.
• Interest and hands-on experience with large language models.
• Knowledge of AI/ML pipelines from development to deployment.
• Strong communication skills for explaining technical topics.

Responsibilities

• Bring inference techniques into production and refine them.
• Design and optimize serving architectures for latency, throughput, and cost.
• Tune the serving stack from frameworks to kernels, profiling for performance.
• Adapt optimization methods across many ML models, focusing on LLMs.
• Profile deployments against latency targets and ensure reliability.
• Collaborate with customer engineering teams to tailor deployments.
• Build and support inference stack software in production using Python/C++.
• Experiment quickly and ship well-tested results.
• Deliver end-to-end optimizations from initial experiments to production.
• Work through ambiguity and make sound trade-off calls.
• Take ownership and accountability in the role.

Skills

Python

C++

LLM inference optimization

Performance profiling

GPU basics

Education

Bachelor's, Master's, or Ph.D. in Computer Science, Engineering, Mathematics, or a related field

Tools

vLLM

SGLang

CUDA

Docker

Kubernetes

Job description

Crusoe is seeking a hands-on engineer to make large language models run faster, cheaper, and more reliably in production. You will own the inference stack end to end, from profiling time and cost to optimizing CUDA kernels and deployment strategies.

You will work with customer teams to tailor deployments, ship production-ready improvements, and drive performance across diverse models using Python as a primary language.