Latest AI JobsBrowse Jobs
Anthropic logo

Performance Engineer, Inference Systems

AnthropicSan Francisco, CA | New York City, NY | Seattle, WAAI Research & Engineering
Apply on company site
Salary
Not listed
Job type
Full-time
Posted
20 May 2026
Status
Open · checked 7d ago

About this role

The Performance Engineer, Inference Systems role at Anthropic involves analyzing and optimizing the inference fleet that serves Claude to millions of users. Based in San Francisco, New York City, or Seattle, this position is ideal for technical professionals who enjoy complex system profiling and ensuring the correctness and reliability of large-scale artificial intelligence infrastructure.

Role Overview

Anthropic's inference fleet relies on a deep, tightly coupled stack covering accelerator kernels, model servers, distributed routing, autoscaling, and capacity management. The Inference System Dynamics team evaluates this entire ecosystem against throughput, latency, reliability, and correctness. This role measures fleet performance against theoretical rooflines, conducts cross-layer investigations, and owns correctness evaluations to ensure Claude delivers accurate outputs across diverse hardware platforms and serving configurations. Candidates will work across multiple areas, tracing latency regressions, modeling performance bottlenecks, and collaborating with cross-functional engineering teams to implement high-impact solutions.

Key Responsibilities

Conduct cross-layer performance investigations focusing on throughput, latency, and reliability to quantify gaps between actual fleet performance and theoretical rooflines.
Own and improve the correctness evaluation pipeline that validates model output quality across various hardware platforms, numerics, and serving configurations.
Build observability metrics, dashboards, and modeling tools to make metrics like throughput, latency, cost, reliability, and correctness legible.
Collaborate closely with kernel, serving, routing, autoscaling, and capacity teams to prioritize and launch impactful optimizations.
Stack-rank optimization opportunities based on impact and effort to focus engineering resources on high-value initiatives.

Requirements

Hands-on performance engineering background covering profiling, roofline analysis, latency or throughput optimization, and root-cause investigations in complex production environments.
Proficiency in Python with the ability to read, instrument, and contribute to large production codebases.
Solid data analysis capabilities using tools such as SQL or pandas to transform raw telemetry into clear insights.
Strong written communication skills to articulate quantitative results and influence cross-functional priorities.
A dedicated interest in correctness as an engineering discipline, encompassing numerics, evaluation design, and regression detection.
A Bachelor's degree or an equivalent combination of education, training, and professional experience in a relevant field.

Nice to Have

Practical experience with machine learning systems, particularly training or inference infrastructure and LLM serving stacks.
Familiarity with GPU, TPU, or accelerator performance concepts including memory bandwidth, kernel overheads, quantization, and collective communication.
Reliability engineering background for high-throughput services such as autoscaling, load balancing, request routing, and tail latency mitigation.
Experience building observability or telemetry systems for distributed architectures.

Skills and Tools

Python
SQL
LLM serving stacks
Pandas
Roofline analysis and performance profiling tools

Salary and Benefits

The annual salary range for this position is $350,000 to $850,000 USD. Anthropic provides competitive compensation, optional equity donation matching, generous vacation and parental leave, and flexible working hours.

About the Company

Anthropic operates as a public benefit corporation dedicated to creating reliable, interpretable, and steerable artificial intelligence systems that remain safe and beneficial for society. The organization functions as a collaborative team of researchers, engineers, and policy experts who approach AI research as an empirical science.

How to Apply

Interested candidates should visit the official Anthropic careers page to submit an application for this position.

Skills and tech stack

PythonSQLLLM

Questions candidates ask

What locations are available for this role?

This position is available in San Francisco, CA, New York City, NY, or Seattle, WA, with a location-based hybrid policy requiring staff to be in the office at least 25% of the time.

What is the salary range for the Performance Engineer position?

The annual salary range is between $350,000 and $850,000 USD.

What programming language is required?

Proficiency in Python is required, along with solid data analysis skills using SQL, pandas, or similar tools.

Does Anthropic offer visa sponsorship?

Yes, Anthropic does sponsor visas and works with an immigration lawyer to support candidates offered a position.

What are the primary responsibilities regarding system performance?

Responsibilities include running cross-layer performance investigations, sizing performance gaps against theoretical rooflines, improving correctness evaluation pipelines, and building observability dashboards.

Ready to apply?

The application is on the official Anthropic careers page.

Apply on company site