Research Engineer, AI Monitoring
Core
Design, train, and evaluate real-time classifiers to detect unsafe actions by internal agents and reinforcement learning systems.
Role type
Senior IC research engineer (AI safety & monitoring)
Builds
Real-time classifiers, datasets, labeling pipelines, and harm taxonomies for agent safety
Domain
AI safety, agent control, reinforcement learning, large language model monitoring
Deliverable
production ML models
Required skills
real-time classifier design, dataset construction, labeling pipeline development, adversarial stress-testing, false positive resolution, technical prioritization
Preferred skills
leading ambiguous technical projects, LLM evaluation, rare-event detection systems, low-latency high-availability operations, model fine-tuning, LLM inference serving, security and sandboxing expertise
Technologies
reinforcement learning, large language models, LLM inference/serving infrastructure, training infrastructure
Responsibilities
Design and deploy classifiers in the critical path of agent training and evaluation, build harm taxonomies to measure performance, partner with red-teaming teams to harden classifiers, resolve false positives while maintaining protection, contribute to team direction and investment decisions
Seniority
Senior, hands-on IC with strategic contribution