Description
Job Overview
As a Research Engineer, AI Monitoring at Anthropic, you will play a crucial role in building, testing, deploying, and monitoring classifiers that prevent internal agents from taking hazardous actions. Your work will contribute to a defence-in-depth strategy, combining multiple layers of protection.
Responsibilities
- Design, train, and evaluate real-time classifiers to detect unsafe agent actions.
- Deploy, tune, and operate these classifiers in critical paths, ensuring accuracy, reliability, latency, cost-effectiveness, and failure resilience.
- Develop datasets, labelling pipelines, and evaluations to enhance classifiers and harm taxonomies.
- Collaborate with red teaming colleagues to stress-test classifiers and improve their robustness.
- Work with teams affected by classifier outputs to resolve false positives efficiently.
- Contribute to the team's strategic direction, including project prioritization and investment areas.
Requirements
- Ability to organize and prioritize multiple tasks and communicate effectively with impacted teams.
- Critical review skills for code and experiment results, including AI-generated content.
- Genuine interest in developing safe and beneficial AI systems.
Preferred Qualifications
- Experience leading complex technical projects and guiding other engineers or researchers.
- Background in evaluating large language models, developing classifiers, or managing high-scale production systems.
- Expertise in detection systems for rare true positives and costly false positives.
- Experience with low-latency, high-availability services and related infrastructure.
Compensation and Benefits
- Annual salary: £375,000 - £640,000 GBP
- Competitive compensation and benefits package, including optional equity donation matching, generous vacation and parental leave, flexible working hours, and a well-equipped office space.
For more fresh new jobs, check out yubhub.co.