Applied Research - Continual Learning (Fully Remote)

Prime Intellect, Inc.; United States; Hybrid; USD 150,000 - 210,000; Full time

Job summary

Prime Intellect, Inc. builds the open superintelligence stack and is hiring for a role owning continual learning systems and online adaptation. You will design experiments to decide what belongs in weights, memory, or prompts and run rigorous tests in production-like settings.

You will develop methods for selective consolidation, replay, adaptation, memory, and online/offline RL, while evaluating retention, transfer, and rollback as models evolve. A fast-moving research team collaborates closely.

Qualifications

• Experience with continual or online learning, RL, meta-learning, or adaptive production ML.
• Strong Python and experimental judgment with clear failure attribution.
• Evidence of exceptional work—publications, open-source systems, or shipped products.

Responsibilities

• Own continual learning systems that convert production traces into training signal and model updates.
• Determine which adaptations belong in weights, memory, skills, or prompts through rigorous experiments.
• Promote gates and human oversight for agents changing in production.

Skills

Continual learning

Online learning

Reinforcement learning

Meta-learning

Adaptive production ML

Python

Experimental judgment

Tools

Python

Job description

• Own continual learning systems that convert production traces into training signal and model updates.
• Determine which adaptations belong in weights, memory, skills, or prompts through rigorous experiments.

What You Will Own:

• Methods for selective consolidation, replay, adaptation, memory, and online/offline RL.
• Evaluations for retention, transfer, regression, and rollback as tasks evolve.
• Promotion gates and human oversight for agents changing in production.

Requirements:

• Experience with continual or online learning, RL, meta-learning, or adaptive production ML.
• Strong Python and experimental judgment with clear failure attribution.
• Evidence of exceptional work—publications, open-source systems, or shipped products.

Prime Intellect

Prime Intellect builds the open superintelligence stack, making frontier AI training and deployment infrastructure available to every ambitious AI team. The company has raised $150M from top investors and is a research-driven, flexible workplace.