Research Engineer, AI Monitoring
Core
Design, train, evaluate, and deploy classifiers to detect unsafe actions by internal AI agents in real-time, blocking dangerous behaviors like sandbox escapes or data exfiltration.
Role type
Senior Research Engineer (AI Safety & Monitoring)
Builds
Real-time safety classifiers, detection pipelines, and evaluation datasets for AI agent safety.
Domain
Artificial Intelligence, AI Safety, Machine Learning, Security
Deliverable
production ML models
Required skills
LLM fine-tuning, classifier design, low-latency high-availability system operations, dataset building, red teaming collaboration, model internals analysis, RL training integration, security background
Preferred skills
Leading ambiguous technical projects, evaluating large language models, building rare-event detection systems, security/sandboxing expertise, scalable oversight research
Technologies
LLM inference/serving infrastructure, RL training environments, model probes, classification systems
Responsibilities
Design and train classifiers to detect unsafe agent actions; Deploy and operate classifiers in the critical path of research and product work; Build datasets and labelling pipelines for classifier improvement; Partner with red teaming to stress-test classifiers; Collaborate with teams to fix false positives without weakening protection; Contribute to team direction on what to build and where to invest
Seniority
Senior, hands-on IC with leadership potential