On-Device AI Foundation Models Researcher

River AI; Palo Alto (CA); On-site; USD 200,000 - 420,000; Full time

Benefits offered by this job

Health, dental & vision insurance

Unlimited PTO

relocation assistance

Job summary

River seeks exceptional researchers to design and train foundation models powering its personal AI. You will push the frontier of deep learning, focusing on architectures that learn continuously, personalize deeply, and run efficiently on local hardware.

You will own the research lifecycle from ideation to scaling large training runs, ensuring AI evolves with users as a true extension of their will. Compensation reflects high performance and impact.

Qualifications

• BS, MS or Ph.D. in CS, ML, Mathematics or equivalent practical industry experience.
• Deep understanding of modern DL architectures (Transformers, diffusion) and training.
• Hands-on with PyTorch or JAX and scalable model training code.
• Ability to take open-ended research problems from theory to scalable implementations.
• Collaborative mindset with bias for action in a fast-paced environment.

Responsibilities

• Design, train, and evaluate foundation models for personalization and local inference.
• Advance continual learning and on-device adaptation without forgetting.
• Develop data-efficient, RLHF-aligned, PEFT techniques for bespoke hardware.
• Collaborate with infrastructure to scale experiments on large compute clusters.

Skills

Deep learning architectures

PyTorch/JAX

Research to implementation

Collaborative mindset

Edge inference

Education

BS/MS/PhD in CS/ML/Math

Job description

River seeks exceptional researchers to design and train foundation models powering its personal AI. You will push the frontier of deep learning, focusing on architectures that learn continuously, personalize deeply, and run efficiently on local hardware.

You will own the research lifecycle from ideation to scaling large training runs, ensuring AI evolves with users as a true extension of their will. Compensation reflects high performance and impact.