FIRST SEENSEP 29, 2026RUNREQ OBSERVATIONDAYS OPEN1555D COMPANY MEDIANTEAM OPEN66AI RESEARCH & ENGINEERING30D TEAM+7NEWLY OBSERVEDTRACKED COMP$500K–$850K /YRDISCLOSED RANGE / ALL TIERSSOURCEVERIFIEDCHECKED 10H AGO
JOB DESCRIPTIONSHORTENED FROM SOURCE
Key responsibilities
4 ITEMS
Study how reinforcement learning training and sampling change with model size, context length, and compute, then improve efficiency.
Develop architectures and reinforcement learning algorithms, run them at frontier scale, and diagnose numerical, algorithmic, or system problems.
Build experimental infrastructure for fast, reproducible comparisons of architecture and algorithm variants at meaningful scale.
Own large reinforcement learning runs from research code through hardware, and investigate instability, divergence, and throughput regressions.
Qualifications needed
4 ITEMS
You hold a bachelor’s degree or equivalent education, training, or experience in a relevant field.
You understand transformer language models, including architecture, training dynamics, and large-scale optimization.
You have trained large models in distributed systems using data, tensor, and pipeline parallelism.
You show original technical work, rigorous experiments, quantitative cost analysis, and strong Python and JAX or PyTorch programming.
EVENT>Anthropic :: Research Engineer / Research Scientist, RL Frontiers first observed 15 days ago