Principal Applied Scientist- Foundation Models, Agents & Decision Intelligence
Core
Define and lead scientific initiatives for foundation models, agentic systems, and decision intelligence to identify risks and model uncertainty in automated decision-making.
Role type
Principal Applied Scientist (Foundation Models, Agents & Decision Intelligence)
Builds
Scalable learning systems, agent trajectories, and AI-driven trust and safety architectures.
Domain
AI Safety, Trust & Safety, Cybersecurity, Fraud Detection
Deliverable
production ML models
Required skills
Probability, statistics, linear algebra, optimization, numerical methods, experimental design, statistical decision theory, foundation/representation learning, behavioral and temporal modeling, anomaly detection, uncertainty modeling, threat modeling, Python, PyTorch, JAX, TensorFlow, tool-using agents, retrieval, reward modeling, trajectory evaluation, graph-structured data processing
Preferred skills
Post-training and evaluation of large-scale models (xxx B param), modeling uncertainty in production decision systems, adversarial ML, multimodal learning, experience with distribution shift and sparse labels
Technologies
PyTorch, JAX, TensorFlow, Python
Responsibilities
Develop methods to model and propagate uncertainty across model cascades and agent trajectories; Translate threat models into data strategies and model architectures; Advance training and evaluation of agents using tools and evidence; Provide technical leadership and mentor scientists; Influence long-term architecture of AI-driven trust and safety systems
Seniority
Principal, hands-on IC with strategic influence