What are the responsibilities and job description for the Principal AI Performance Engineer position at Jobright.ai?
Jobright is an AI-powered career platform that helps job seekers discover the top opportunities in the US. We are NOT a staffing agency. Jobright does not hire directly for these positions. We connect you with verified openings from employers you can trust.
Job Summary:
Crusoe is building the World’s Favorite AI-first Cloud infrastructure company, pioneering AI infrastructure solutions for Fortune 500 companies. As a Principal AI Performance Engineer, you will optimize scalable inference engines to enhance performance, efficiency, and speed, directly impacting Crusoe’s revenue model.
Responsibilities:
• Optimize inference engines – Improve inference performance in engines such as VLLM, ensuring maximum efficiency and scalability.
• Enhance scalable AI infrastructure – Implement optimizations that accelerate AI inference, directly impacting Crusoe’s efficiency and revenue generation.
• Develop CUDA kernels – Write and deploy CUDA kernels to optimize deep learning workloads, improving computational performance.
• Conduct performance analysis – Profile and analyze training and inference workloads to identify and resolve bottlenecks.
• Engage with the AI research community – Track developments in scalable inference, contribute to open-source projects, and publish research to advance the field.
• Improve onboarding and documentation – Enhance internal documentation and tooling standards to streamline team workflows and training.
• Collaborate cross-functionally – Work closely with AI researchers, engineers, and infrastructure teams to develop cutting-edge solutions.
Qualifications:
Required:
• Expertise in CUDA or OpenCL – Demonstrated experience developing CUDA kernels or equivalent technologies.
• Proficiency in Python – Strong programming skills, particularly in Python, for AI and performance optimization tasks.
• Experience with deep learning frameworks – Hands-on knowledge of training infrastructure such as PyTorch or TensorFlow.
• Strong understanding of CPU & GPU architecture – Ability to analyze and optimize performance at the hardware level.
Preferred:
• Zero-to-Hero mindset – Experience taking a project from initial concept to full implementation.
• Experience with vector instructions – Understanding of SIMD, AVX, or similar vector processing techniques.
• Graphics shader knowledge – Background in graphics shaders as a proxy for CUDA expertise.
Company:
Crusoe is the industry’s first vertically integrated, purpose-built AI cloud platform. Founded in 2018, the company is headquartered in Denver, Colorado, USA, with a team of 501-1000 employees. The company is currently Late Stage. Crusoe has a track record of offering H1B sponsorships.
