Staff Research Engineer, Multi-Agent Scaling
About this role
This role addresses the challenge of understanding how large teams of AI agents scale to solve complex problems, such as rewriting large codebases or formalizing mathematics. The team is focused on building the platform and evaluation systems that allow Anthropic researchers to effectively run, measure, and interpret experiments involving these large agent teams. This position sits at the intersection of research and engineering, requiring the ability to move quickly from a high-level research question to a running experiment and then rigorously analyze the outcomes. The core objective is to identify performance, cost, and coordination bottlenecks as team size, compute, and task complexity increase, ensuring that these large-scale systems remain reliable and efficient.
The successful candidate will be responsible for designing large-scale experiments, building the underlying systems to support them, and developing the necessary tooling and metrics for researchers to clearly understand agent team activities. This is a generalist role that demands both deep technical engineering skills and a strong research mindset, requiring an ability to think quantitatively about complex systems and navigate ambiguity.
This role is ideal for someone who enjoys digging into complex systems, building robust infrastructure, and translating empirical results into actionable insights for the broader research community.
What you will do
- Design, execute, and interpret large-scale experiments focused on agent teams.
- Investigate how performance and efficiency change relative to team size, compute budget, and task horizon.
- Identify the bottlenecks that limit the scaling of agent teams.
- Build and scale the systems necessary to run very large agent teams reliably.
- Debug failures that manifest only at scale.
- Design evaluations specifically for long-horizon problems to ensure result trustworthiness.
- Develop tooling and metrics that enable researchers to monitor large agent team operations.
- Collaborate with research teams to allow them to run their own experiments on the platform and communicate findings clearly.
What you need
- Possess significant experience in software engineering, machine learning, or research engineering.
- Must have owned substantial end-to-end projects, such as a large system, evaluation, benchmark, agent product, or research project.
- Demonstrate a genuine enjoyment of both research and engineering work.
- The candidate must be able to think quantitatively about complex systems and exercise caution when interpreting data.
- Ability to work effectively from vague questions rather than detailed specifications.
- Must be results-oriented with a preference for flexibility and impact.
- Requires clear written and verbal communication skills to convey complex findings.
- A background in operations research, statistics, economics, physics, quantitative finance, or a similar field that models and optimizes complex systems is beneficial.
Nice to have
- Experience building or operating large-scale distributed systems, such as schedulers, sandboxed code execution, or inference and RL infrastructure
- Experience building evaluations, benchmarks, or harnesses for LLMs or agents
- Experience building complex agentic systems that utilize LLMs
- Experience with scaling laws or other large-scale empirical research
Benefits
- $500,000 — $850,000 USD Annual Salary
Skills and tech stack
Questions candidates ask
What is the remote policy?
Currently, all staff are expected to be in one of Anthropic's offices at least 25% of the time. However, some roles may require more time spent in the offices.
What is the compensation range for this role?
The annual compensation range for this role is listed between $500,000 and $850,000 USD.
What is the primary tech stack involved?
The role involves working with LLMs and multi-agent systems. Nice-to-have experience includes building systems like schedulers, sandboxed code execution, or inference and RL infrastructure.
What is the hiring process regarding visa sponsorship?
Anthropic does sponsor visas, but they are not able to sponsor every role and every candidate. If an offer is made, they will make every reasonable effort to secure a visa and retain an immigration lawyer to assist.