Member of the Technical Staff, Pretraining

📋 Description

• Advance the foundation model's architecture and training.
• Design training objectives and scale pretraining.
• Run distributed pretraining on multi-GPU clusters.
• Evaluate what the model learns and its biological reasoning.
• Own research-to-training pipeline end-to-end.

🎯 Requirements

• PhD in CS/ML/physics/math or related field; 2+ years postdoc or industry research.
• Or Bachelor's/Master's with 5+ years hands-on representation learning and model pretraining.
• Strong publication record at NeurIPS/ICML/ICLR.
• Hands-on experience pretraining large models on diverse data.
• Proficient in Python and PyTorch; distributed multi-GPU infrastructure.
• Own full research-to-training pipeline; ship models.
• Production-quality code; version control and code review.
• Rigorous experimentalist; designs evaluations and tracks experiments.

🎁 Benefits

• Healthy, feedback-focused environment with constructive feedback and growth.
• Ownership of day-to-day management; milestones-focused work.
• Excellent medical, dental, and vision coverage.
• Culture values: ownership, excellence, practicality, honesty, and fun.
• Opportunities to contribute to open-source ML projects.