Periodic Labs is an AI and physical sciences lab focused on building state-of-the-art models to enable novel scientific discoveries. The company is well funded, growing rapidly, and values team members who take ownership and solve problems without bureaucracy. The team embraces learning new tools and science to advance their mission. The Research Engineer - Midtraining role involves training frontier models to be highly knowledgeable scientific experts foundational for reinforcement learning. Responsibilities include developing methods for synthetic data generation, distillation, and continual learning at scale; collaborating with RL researchers, physicists, and chemists to create evaluations guiding scientific data curation; working with supercompute engineers to scale compute-efficient LLM training across thousands of GPUs; and building high-performance tools to investigate how data shapes intelligence. Ideal candidates have experience with:

• Training LLMs on curated mixes of trillions of tokens
• Calculating scaling laws and compute-optimal hyperparameters
• Generating billions of tokens of high-quality synthetic data
• Building evaluations that correlate with downstream task performance