Research Engineer, AI Computer Use & Vision

Anthropic; United States; Hybrid; USD 500,000 - 850,000; Full time

Benefits offered by this job

Competitive compensation

Optional equity donation matching

Generous vacation and parental leave

Flexible working hours

Office space

Job summary

Anthropic in the United States is seeking a Research Engineer for the Computer Use team to advance Claude's ability to see, use, and understand real software. You will design experiments and build evaluation frameworks to improve model performance on complex computer tasks.

You will collaborate across model training and product teams to deploy research into production, leveraging Python, ML training, and safety-focused approaches in a hybrid work environment.

Qualifications

• Bachelor's degree or higher in a relevant field.
• Experience in software engineering and Python.
• Experience training, fine-tuning, or evaluating ML models.
• Strong communication skills and collaboration.

Responsibilities

• Design and run experiments to improve Claude's perception and agentic capabilities.
• Develop robust evaluation frameworks for measuring model performance on complex tasks.
• Build and improve computer use and vision RL training environments.
• Create pipelines and tools to test and validate RL environments.
• Collaborate with teams across model training and infrastructure to improve production setup.
• Partner with product teams to bring research advances into production.

Skills

Python

ML model training

Evaluation frameworks

Communication

Collaborative work

Education

Bachelor's degree

Tools

PyTorch

RL environments

ML infrastructure

Job description

Anthropic in the United States is seeking a Research Engineer for the Computer Use team to advance Claude's ability to see, use, and understand real software. You will design experiments and build evaluation frameworks to improve model performance on complex computer tasks.

You will collaborate across model training and product teams to deploy research into production, leveraging Python, ML training, and safety-focused approaches in a hybrid work environment.