Research Engineer or Research Scientist RL Frontiers

Skills

About the Role

You will study reinforcement learning at increasing model, context, and compute scale. You will develop architectures and algorithms, build reproducible experimental infrastructure, diagnose training issues, and improve the performance and cost of large-scale training runs.

Requirements

• Modern transformer language model architecture, training dynamics, and large-scale optimization knowledge
• Experience training large distributed models using data, tensor, and pipeline parallelism
• Track record of original technical work in machine learning training or systems
• Ability to design rigorous large-scale experiments with baselines and ablations
• Ability to reason quantitatively about compute, memory, and communication costs
• Python and JAX or PyTorch programming skills

Responsibilities

• Study how reinforcement learning training and sampling scale with model size, context length, and compute
• Develop model architectures and reinforcement learning algorithms for frontier-scale execution
• Scale promising small-scale results and diagnose numerical, algorithmic, and systemic differences
• Build reproducible experimental infrastructure for architecture and algorithm comparisons
• Own end-to-end performance of large reinforcement learning runs
• Build performance and cost models for architecture and algorithm changes
• Investigate and trace training instabilities, divergence, and throughput regressions to root causes

Benefits

• Optional equity donation matching
• Generous vacation
• Parental leave
• Flexible working hours