Performance Engineer (ML Systems)

About the role

Anthropic is looking for a Performance Engineer to tackle the unique systems challenges that arise when running large‑scale machine‑learning workloads. You will work closely with researchers and engineers to make our distributed AI infrastructure faster, more reliable, and easier to scale.

Key responsibilities

• Identify bottlenecks in large‑scale ML pipelines and design solutions that improve throughput and latency.
• Develop low‑latency, high‑throughput sampling and inference kernels for transformer models.
• Implement custom load‑balancing and fault‑tolerant algorithms for complex network topologies.
• Build quantitative performance models and conduct systematic benchmarking.
• Debug kernel‑level network latency spikes in containerized environments.

Required profile

• Significant software engineering or machine‑learning experience at supercomputing scale.
• Results‑oriented mindset with a bias toward impact and flexibility.
• Willingness to work across disciplines, including pair programming and cross‑team collaboration.
• Interest in the societal implications of AI technology.

Required skills

• GPU/Accelerator programming
• ML framework internals
• OS internals
• Transformer language‑modeling expertise
• Distributed systems design
• Load‑balancing algorithm development
• Performance modeling and quantitative analysis

What we offer

• Competitive annual compensation ranging from $280,000 to $850,000 USD.
• Hybrid work model with at least 25% onsite presence in our San Francisco office.
• Visa sponsorship where possible.
• Opportunity to work on cutting‑edge AI research and engineering.