Full-time position
Applied Researcher - Continual Learning
Prime Intellect
- San Francisco, United States
- Machine Learning, Artificial Intelligence and
About
Build systems that learn from deployed agents’ trajectories, corrections and failures. Compare streaming reinforcement learning, batched post-training and distillation. Determine which information belongs in model weights and which belongs in persistent memory. Design experiments that measure retention, transfer and regression across sequences of tasks.
Requirements
Experience in continual or online learning, reinforcement learning, meta-learning, memory systems or related evaluation. Hands-on model training and evaluation, experimental design skills and strong Python.
Topics
- Continual Learning, search World Wide
- transfer, search World Wide
- deployed agents, search World Wide
- trajectories, search World Wide
- corrections, search World Wide
- streaming reinforcement learning, search World Wide
- batched post-training, search World Wide
- distillation, search World Wide
- model weights, search World Wide
- persistent memory, search World Wide
- retention, search World Wide
Source: Prime Intellect