Research Engineer, Safeguards Labs
Core
Define and execute research agenda to develop novel safety methods, detect misuse of Claude, and build classifiers for abuse patterns.
Role type
Research Engineer (AI Safety & Safeguards)
Builds
Offline analysis systems, detection classifiers, and prototypes for real-time safeguards.
Domain
Artificial Intelligence, Large Language Models, AI Safety, Trust & Safety
Deliverable
production ML models
Required skills
Python, large dataset analysis, independent project scoping, LLM operation familiarity (sampling/prompting/training), experimental design
Preferred skills
Machine learning model training, evaluation methodologies for language models, agentic environment evaluation, red teaming/jailbreak research, interpretability methods, prototype-to-production transfer
Technologies
Python
Responsibilities
Lead research projects on detecting misuse and malicious accounts; Design and run offline analyses over model usage data; Develop and iterate on prototypes for real-time safeguards; Contribute to research on detecting abusive behavior in chat/agentive workflows; Build evaluations for measuring safeguard effectiveness; Write findings to inform decisions across teams.
