Periodic Labs
Research Engineer - Midtraining
About this role
Join a frontier AI company to develop scientifically-grounded language models by curating data, building evaluations, and running large-scale training experiments. You'll enhance model reasoning across materials, energy, and physical sciences while collaborating with researchers and infrastructure teams.
What you'll do
- Identify, process, and curate novel scientific data sources for training at scale
- Generate high-quality synthetic data to strengthen model scientific reasoning
- Build evaluations that predict downstream scientific task performance
- Apply distillation techniques and optimization methods to improve model capabilities
- Design and execute large-scale training experiments across thousands of GPUs
- Develop tools for analyzing how data choices impact model intelligence
What they're looking for
- LLM training on large curated token mixes
- Mid-training and pre-training at scale
- Evaluation design and benchmarking
- Self-distillation and on-policy distillation
- Scaling laws and compute-optimal hyperparameter tuning
- Distributed training optimization and reliability
- AI for science and domain-specific datasets
- Data curation and synthetic data generation
Benefits
- Equity compensation
- Visa sponsorship available
- Work on frontier AI for scientific discovery
- Collaborate with physicists and chemists
- Access to large-scale GPU infrastructure
- Rapid growth environment with world-class investors
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Periodic Labs
Periodic Labs builds AI systems and autonomous lab infrastructure to accelerate materials discovery and scientific research, integrating machine learning with physical lab workflows and quantum simulation. The company is hiring for infrastructure engineers, product engineers, research engineers, and deployment specialists who bridge AI models with materials scientists and lab automation systems.
View all jobs at Periodic LabsLikely interview questions
- Describe your experience training LLMs on curated token mixes at scale and what challenges you encountered.
- Walk us through how you've applied self-distillation or on-policy distillation in a production training pipeline.