Staff / Principal Research Scientist

Skills

About the Role

You'll dive into unclear, ambiguous problems and make them clear, framing the right questions and designing experiments to find answers. You'll work on foundation models, evaluation systems, and frontier topics like multimodal models, agents, tool use, and world models. You'll treat evaluation as a first-class research product rather than a checkbox, and you'll own your research end-to-end—from framing the question, to running experiments, to writing up and shipping the result. You'll collaborate in person with a flat, fast-moving team, question existing approaches when you see a better path, and prioritize impact over purely academic output, while still being supported in sharing work that advances the field.

Requirements

• Experience with foundation models, including training, new architectures, reinforcement learning, reward modeling, and scaling
• Experience with evaluation, including benchmarks, evaluation loops, quality measurement, LLM-as-judge, and failure analysis
• Experience with frontier topics including multimodal models, agents, tool use, test-time compute, and world models
• Published research at ICML, ICLR, NeurIPS, EMNLP, ACL, or AAAI
• PhD in ML/NLP or equivalent practical experience
• Public work such as non-trivial AI side projects, interdisciplinary experiments, or open-source contributions
• Full-stack research ownership from framing questions to shipping results

Responsibilities

• Frame research questions and design experiments
• Run experiments and analyze results
• Write up and ship research results
• Build and improve foundation models including training, architectures, reinforcement learning, reward modeling, and scaling
• Design evaluation benchmarks, evaluation loops, and quality measurement systems
• Perform failure analysis on models and systems
• Work on frontier topics such as multimodal models, agents, tool use, test-time compute, and world models

Benefits

• Relocation assistance
• Equity
• Bonus