Research Scientist: RL & Robotics for Human‑AI Collaboration

Frisson Labs; San Francisco (CA); On-site; USD 140,000 - 190,000; Full time

Job summary

Frisson Labs in San Francisco develops learning-based companions and embodied AI. This role focuses on creating policies that learn from demonstrations and improve through experience, combining RL, imitation learning, and robotics.

We seek a researcher who will build simulation environments and evaluation tasks, publish findings, and contribute to scalable training pipelines. A PhD and strong publication record are essential.

Qualifications

• PhD required or equivalent record of original research.
• Strong publication record in major ML/robotics venues.
• Deep understanding of RL and sequential decision-making.
• Proficiency in Python and PyTorch for experiments.
• Experience with simulation, embodied learning and robotics.

Responsibilities

• Develop policies that combine imitation learning with offline and online reinforcement learning, using demonstrations and experience.
• Build simulation environments, training curricula, and evaluation tasks for control, exploration, planning and human cooperation.
• Research world models and hierarchical policies linking observations to actions under partial observability.
• Study policy generalization across environments, tasks, and interfaces including sim-to-real adaptation.
• Design rewards, collect training data, and run controlled experiments separating task performance from data/evaluation changes.
• Implement efficient training and inference systems and connect policies to perception, memory, and conversational behavior.

Skills

PhD in robotics, ML, CS

Publications at NeurIPS/ICLR/ICML/CoRL

Reinforcement learning & sequential DM

Python & PyTorch

Simulation & robotic control

Education

PhD in robotics, ML, CS, or related field

Tools

PyTorch

Job description

Frisson Labs in San Francisco develops learning-based companions and embodied AI. This role focuses on creating policies that learn from demonstrations and improve through experience, combining RL, imitation learning, and robotics.

We seek a researcher who will build simulation environments and evaluation tasks, publish findings, and contribute to scalable training pipelines. A PhD and strong publication record are essential.