Research Scientist/Research Engineer, Midtraining

What you'll do

• Identify, process, and curate scientific training data; generate synthetic data to fill knowledge and reasoning gaps.
• Build evaluations tied to downstream scientific task performance and develop methods such as self-distillation and on-policy distillation.
• Design and run large-scale training experiments, scale across thousands of GPUs, and build tools to study how data choices affect model intelligence.

What they ask for

• Experience training LLMs on curated mixes of trillions of tokens.
• Experience on a dedicated evaluations team supporting a large production training run.
• Hands-on use of self-distillation, on-policy distillation, or similar methods in a real training pipeline.
• Experience with scaling laws and compute-optimal hyperparameters; comfort across data, evaluations, and training infrastructure.

Nice to have

• Experience optimizing throughput and reliability for large-scale distributed training runs.
• Background in AI for science or specialized scientific data; experience creating evaluations or synthetic data for non-verifiable tasks and tracking live-run performance.

Pay

$250,000–$350,000 USD base salary; equity amount not stated

Tools

LLMs

AI focus

Train frontier models for scientific knowledge and reasoning, including data curation and synthesis, evaluations, distillation, and large-scale training.