🛠 Tech Stack

💼 About This Role

You'll develop a rigorous theory of latent internal structure in neural networks at a mission-driven startup. Your work will directly inform AI safety by uncovering belief geometries and generalization mechanisms. You'll collaborate across fields like physics, neuroscience, and machine learning.

🎯 What You'll Do

• Design and run experiments on large language models
• Develop mathematical theory and write proofs
• Implement unsupervised methods for belief geometry discovery
• Collaborate on red teaming and theory validation

📋 Requirements

• PhD or equivalent deep experience in physics, mathematics, neuroscience, or machine learning
• Strong background in dynamical systems, probability, or information theory
• Proficiency in Python and deep learning frameworks
• Experience with transformers or neural network internals

✨ Nice to Have

• Familiarity with AI safety research
• Experience with biological neural data
• Published research in mechanistic interpretability

🎁 Benefits & Perks

• 📈 Ownership over research directions
• 🤝 Small team collaboration
• 🧠 Cutting-edge AI safety work
• 🌐 Interdisciplinary research environment