Performance Engineer, Inference Engine - Flexible Hours

Anthropic; San Francisco, New York (CA, NY); On-site; USD 350,000 - 850,000; Full time

Job summary

Anthropic is seeking a Performance Engineer for the Inference Engine to optimize throughput, latency, and reliability across accelerator and cloud platforms, enabling Claude-scale workloads.

You will model hardware bounds, tune memory, and coordinate host-device interactions, applying deep systems programming to high-performance distributed systems. Collaboration with research, safety, and platform teams is essential.

Qualifications

• LLM inference concepts including prefill and decode on accelerator compute, memory, interconnect
• Demonstrated rapid learning in deep, unfamiliar systems and shipped consequential changes quickly
• Strong systems programming with emphasis on code quality and tests
• Analytical about performance: observe, profile, hypothesize, test, and measure
• Low ego: ask questions, take feedback, and assist beyond job scope
• Enjoy pair programming and collaboration across teams

Skills

LLM inference

Fast learner

Systems programming

Performance analysis

Team collaboration

Pair programming

Tools

Rust

Job description

Anthropic is seeking a Performance Engineer for the Inference Engine to optimize throughput, latency, and reliability across accelerator and cloud platforms, enabling Claude-scale workloads.

You will model hardware bounds, tune memory, and coordinate host-device interactions, applying deep systems programming to high-performance distributed systems. Collaboration with research, safety, and platform teams is essential.