ML Systems Manager: EC2 Nitro Performance

Amazon Web Services (AWS); Seattle (WA); On-site; USD 185,000 - 250,000; Full time

Benefits offered by this job

Health insurance

RSUs

401(k) matching

Paid time off

Parental leave

Job summary

Amazon Development Center U.S., Inc. is seeking a leader to drive a pivotal ML performance engineering group within EC2 Nitro ML Systems.

You will build and lead a team focused on optimizing performance across diverse workloads including LLMs and multimodal models, influencing product decisions with comprehensive performance data. The role spans low‑level CUDA optimization through frameworks to serving layers, translating insights into platform design, and establishing scalable benchmarking that

Qualifications

• 3+ years of engineering team management experience.
• 7+ years working directly within engineering teams.
• 3+ years designing or architecting scalable systems.
• Knowledge of end-to-end software/hardware/network lifecycle.
• Experience partnering with product or program management teams.

Responsibilities

• Lead design, implementation, and delivery of foundational ML performance measurement infrastructure across accelerator platforms.
• Build and mentor a high-performing ML performance engineering team.
• Influence future platform design by feeding insights from research and customer workloads into launches.
• Develop regression coverage across frameworks, firmware, drivers, and networking components.
• Scale performance engineering from LLMs to multimodal models and new domains.

Skills

Team management

Engineering leadership

System design

Software lifecycle

PM collaboration

Job description

Amazon Development Center U.S., Inc. is seeking a leader to drive a pivotal ML performance engineering group within EC2 Nitro ML Systems.

You will build and lead a team focused on optimizing performance across diverse workloads including LLMs and multimodal models, influencing product decisions with comprehensive performance data. The role spans low‑level CUDA optimization through frameworks to serving layers, translating insights into platform design, and establishing scalable benchmarking that