Ir para o conteúdo
AI.info

jobs

Research Engineer, Post-Training Model Evaluations

Anthropic's production models undergo post-training to improve capabilities, alignment, and safety. The Post-Training Model Evaluations team works to provide a clear and trustworthy view of model…

Em inglês

Company
Anthropic
Location
San Francisco, CA; Seattle, WA
Status
Closed
Posted
2026-09-29 03:33 UTC

Anthropic's Post-Training Model Evaluations team aims to give a trustworthy view of model quality during post-training runs. The research engineer sets evaluation strategy for production Claude models, tunes the evaluation mix, monitors evaluations during live training, advises post-training teams on methodology, and investigates regressions and unexpected results. It also builds dashboards, alerts and reports for researchers and leadership. The posting asks for experience designing, running and analyzing evaluations of machine-learning models at scale, strong Python, comfort with production systems, and statistics or experimental-design background.

Original job posting