Research Member of Technical Staff – Post‑training & Robot Learning

About the role

Rhoda AI is building generalist intelligent robots and needs researchers to adapt large video models to real‑world robot tasks. You will lead post‑training efforts that turn internet‑scale pretrained models into reliable robot policies with minimal task‑specific data.

Key responsibilities

• Design and implement reinforcement‑learning pipelines (e.g., PPO, GRPO) to improve robot policies beyond imitation learning.
• Develop supervised fine‑tuning, preference optimization, and behavioral alignment pipelines on robot‑collected demonstrations.
• Build evaluation frameworks for task success, generalization, and failure‑mode analysis on real hardware.
• Iterate quickly between simulation and real‑robot testing, collaborating with the pre‑training team to identify missing capabilities.

Required profile

• Hands‑on experience with robot systems, policy learning, and autonomous‑systems research or industry work.
• Strong understanding of imitation learning, behavior cloning, and how RL builds on them.
• Ability to diagnose policy failures, reason about distribution shift, and iterate on data and training strategies.
• Comfort with fast‑changing research priorities and ambiguous problems.

Required skills

• Python programming and PyTorch.
• Reinforcement‑learning algorithms (PPO, GRPO) and reward design.
• Robot middleware such as ROS/ROS2.
• Simulation environments (MuJoCo, Isaac Sim, Genesis).
• Experience with video generation models and translating predictions to robot actions.