To drive the advancement of continual learning systems, the full-time Continual Learning Researcher will build and evaluate methodologies for turning agent trajectories and model updates into new tasks while collaborating with external partners, working remotely or from locations such as San Francisco, London, or New York.
Key responsibilities:
Develop methods for selective consolidation, replay, adaptation, and reinforcement learning
Build task-generation and training pipelines with reproducible data to trace regressions
Conduct experiments to measure skill retention, transfer, and regression across changing tasks
Required qualifications:
Experience in continual or online learning, reinforcement learning, or adaptive production machine learning
Hands-on experience with model-training and evaluation experimental loops
Strong Python skills with the ability to create artifacts for research and engineering use
Demonstrated experimental judgment with a focus on meaningful baselines and clear failure attribution
Ability to manage ambiguous problems and communicate effectively with partners and teammates