Principal AI Research Engineer – Reinforcement Learning

About the role

Reflex Robotics is seeking a senior researcher to design and implement robust on‑policy reinforcement‑learning policies for its AI‑powered humanoid robots. You will work directly with hardware teams to translate cutting‑edge algorithms into reliable real‑world performance.

Key responsibilities

• Develop and fine‑tune on‑policy RL algorithms (e.g., SAC, DDPG) to achieve near‑perfect success rates on deployed robots.
• Contribute to sample‑efficient RL research, extending methods such as DreamerV3 and MuZero for robotics.
• Integrate RL policies with robot hardware, validating performance on tasks like quadruped walking and drone racing.
• Collaborate with cross‑functional teams to ensure policies meet strict reliability and cost targets.

Required profile

• Proven experience re‑implementing core RL algorithms from scratch and debugging unstable gradients.
• Track record of publishing or delivering intellectual contributions to sample‑efficient RL.
• Hands‑on experience shipping on‑policy RL on real hardware.
• Passion for high‑impact robotics applications and willingness to own end‑to‑end solutions.

Required skills

• SAC (Soft Actor‑Critic)
• DDPG (Deep Deterministic Policy Gradient)
• DreamerV3
• MuZero
• On‑policy reinforcement learning
• Hardware‑in‑the‑loop RL deployment (e.g., quadruped, drone)