Staff Software Engineer — RL Infrastructure & Platforms

Anthropic; New York, Seattle, San Francisco (NY, WA, CA); On-site; USD 140,000 - 180,000; Full time

Job summary

Anthropic is seeking an engineer to help shape the engineering foundations for reinforcement learning-enabled coding capabilities. You will design widely used APIs and frameworks, embed with research teams on a rotational basis, and maintain production RL systems with a focus on reliability and diagnostics.

You will collaborate with researchers and engineers to build scalable, maintainable systems, improving interfaces and abstractions while navigating complex research codebases and hidden

Qualifications

• Deep expertise in Python, including typing and async/concurrency patterns.
• Experience designing intuitive APIs or frameworks adopted by teams.
• Experience working in large, evolving or research-style codebases.
• Ability to anticipate failure modes and enforce reliable designs via testing.

Responsibilities

• Design widely used APIs, frameworks, and abstractions for engineers and researchers.
• Embed with research teams on a rotational basis and transfer ownership to teams.
• Improve reliability of research codebases without slowing down progress.
• Anticipate silent failures and reduce surface area with type safety and invariants.
• Contribute to production RL system reliability, monitoring, and triage tooling.
• Define engineering standards and mentor researchers and engineers.

Skills

Python

API design

Large codebases

System design

Written communication

Ambiguity handling

Job description

Anthropic is seeking an engineer to help shape the engineering foundations for reinforcement learning-enabled coding capabilities. You will design widely used APIs and frameworks, embed with research teams on a rotational basis, and maintain production RL systems with a focus on reliability and diagnostics.

You will collaborate with researchers and engineers to build scalable, maintainable systems, improving interfaces and abstractions while navigating complex research codebases and hidden