[P] Compute Efficiency Engineer

Anthropic•San Francisco, CA

19h•$1 - $2•Hybrid

About The Position

Anthropic’s mission is to create reliable, interpretable, and steerable AI systems. We want AI to be safe and beneficial for our users and for society as a whole. Our team is a quickly growing group of committed researchers, engineers, policy experts, and business leaders working together to build beneficial AI systems. At Anthropic, we are building some of the most complex and large-scale AI infrastructure in the world. As that infrastructure scales rapidly, so does the imperative to optimize how we use it. As an Infrastructure Optimization & Efficiency Engineer on the Capacity team, you will play a central role in making our systems more performant, cost-effective, and sustainable—without compromising reliability or latency. You will work across the full infrastructure stack, from cloud platforms and networking to application-level performance, and will bridge the gap between high-level research needs and low-level hardware constraints to build the most efficient AI infrastructure in the world. You will help with building the telemetry, cost attribution, and optimization frameworks that ensure every dollar of our infrastructure investment delivers maximum value. This is a high-impact, cross-functional role at the intersection of systems engineering, financial optimization, and AI infrastructure.

Requirements

Have 6+ years of relevant industry experience, 1+ year leading large scale, complex projects or teams as an engineer or tech lead
Deep expertise in distributed systems at scale, with a strong focus on infrastructure reliability, scalability, and continuous improvement.
Strong proficiency in at least one programming language (e.g., Python, Rust, Go, Java)
Hands-on experience with cloud infrastructure, including Kubernetes, Infrastructure as Code, and major cloud providers such as AWS or GCP.
Experience optimizing end-to-end performance of distributed systems, including workload right-sizing and resource utilization tuning.
You possess a deep curiosity for how things work under the hood and have a proven ability to work independently to solve opaque performance issues
Experience designing or working with performance and utilization monitoring tools in large-scale, distributed environments.
Strong problem-solving skills with the ability to work independently and navigate ambiguity.
Excellent communication and collaboration skills—you will work closely with internal and external stakeholders to build consensus and drive projects forward.

Nice To Haves

Experience with machine learning infrastructure workloads as well as associated networking technologies like NCCL.
Low level systems experience, for example linux kernel tuning and eBPF
Quickly understanding systems design tradeoffs, keeping track of rapidly evolving software systems
Published work in performance optimization and scaling distributed systems

Responsibilities

Build and evolve telemetry and monitoring systems to provide deep visibility into infrastructure performance, utilization, and costs across our cloud and datacenter fleets.
Design and implement cost attribution frameworks for our multi-tenant infrastructure, enabling teams to understand and optimize their resource consumption.
Identify and resolve performance bottlenecks and capacity hotspots through deep analysis of distributed systems at scale.
Partner closely with cloud service providers and internal stakeholders to optimize cluster configurations, workload placement, and resource utilization across AI training and inference workloads—including large-scale clusters spanning thousands to hundreds of thousands of machines.
Develop and champion engineering practices around efficiency, driving a culture of performance awareness and cost-conscious design across Anthropic.
Collaborate with research and product teams to deeply understand their infrastructure needs, and design solutions that balance performance with cost efficiency.
Drive architectural improvements and code-level optimizations across multiple services and platforms to deliver measurable utilization and performance gains.