Research Engineer, Computer Use

Anthropic•Seattle, WA

1d•$500,000 - $850,000•Hybrid

About The Position

The Computer Use team focuses on teaching Claude to see, use, and understand computer interfaces. As a Research Engineer on the team, you'll work on advancing our models' ability to reliably and safely operate real software. We're looking for someone who's genuinely excited about both the research and the product sides of computer use. Your work will translate directly into model improvements in our own and our customers' products. You can try Claude's computer use capabilities today through the Claude in Chrome extension and Claude Cowork.

Requirements

Software engineering experience and proficiency in Python
Experience training, fine-tuning, or evaluating machine learning models
Strong communication skills and a collaborative working style
Care about the societal impacts and safety of your work

Nice To Haves

Experience training models for computer use or other agentic capabilities
Experience with reinforcement learning, particularly in long-horizon or sparse-reward settings
Familiarity with multimodal model training
Experience building evaluations or benchmarks for agentic systems
Experience building reinforcement learning environments, simulation systems, or large-scale ML infrastructure
Experience working closely with product teams to drive model improvements

Responsibilities

Design and run experiments to improve Claude's perception and agentic capabilities
Develop robust, reliable evaluation frameworks for measuring our models' ability to complete complex computer tasks
Build and improve computer use and vision reinforcement learning training environments
Create pipelines and tools to test and validate complex RL environments
Collaborate with teams across the model training and infrastructure stack to improve our production training setup
Partner with product teams to bring research advances into production