Research Engineer / Research Scientist (RL Frontiers)

Join Anthropic, a leading AI safety and research company, as a Research Engineer/Research Scientist focused on Reinforcement Learning (RL). In this role, you will work at the intersection of research and engineering, developing next-generation architectures and RL algorithms, and scaling them to frontier levels. You will build the systems that drive research velocity, own the performance of large RL runs, and investigate training dynamics at scale. This position offers a comprehensive benefits package, including health insurance, paid parental leave, flexible time off, and competitive salary and equity packages.

• Studying how reinforcement learning (RL) training and sampling scale with model size, context length, and compute, and finding the algorithmic and systems changes that keep scaling efficient.
• Developing next-generation model architectures and RL algorithms, and making them run efficiently at frontier scale, including taking promising small-scale results to frontier-scale runs.
• Building the experimental infrastructure that sets research velocity, enabling fast, reproducible comparisons of architecture and algorithm variants at meaningful scale.

• A track record of original technical work in ML training or systems, such as new methods, architectures, or optimizations, demonstrated through research, open-source, or production impact
• Deep familiarity with modern transformer language models, including their architecture, training dynamics, and the behavior of large-scale optimization
• Ability to design rigorous experiments at scale, including baselines, ablations, and enough statistical care to trust a result that costs real compute
• Ability to reason quantitatively about the compute, memory, and communication costs of a model or algorithm
• Strong programming skills in Python and JAX or PyTorch, and comfort reading and changing code at every layer of the stack
• Hands-on experience training large models in a distributed setting, including the tradeoffs between data, tensor, and pipeline parallelism
• Minimum education: Bachelor’s degree or an equivalent combination of education, training, and/or experience
• We encourage you to apply even if you do not believe you meet every single qualification. Not all strong candidates will meet every single qualification as listed
• Research experience in reinforcement learning, optimization, or large-scale training, published or otherwise
• Experience developing RL algorithms for language models
• Experience with scaling laws or other quantitative models of training efficiency
• Experience designing or modifying transformer architectures beyond standard configurations
• Experience scaling training to large fleets of accelerators and debugging the problems that only appear at scale
• Familiarity with how GPU or TPU performance characteristics shape architecture and algorithm choices
• Deep understanding of numerics in large-scale training, including low-precision formats and sources of instability
• Experience with C++ or Rust