Moonlake

Lead the development of reinforcement-learning and post-training systems that improve multimodal and code-generating agents. This hands-on research and engineering role spans trajectory learning, rewards, evaluation, and scalable distributed training for embodied AI.

Design and train general embodied AI agents that perceive, reason, and act in simulated worlds. This is an on-site role with Moonlake’s in-person team in San Mateo.

Optimize Moonlake’s machine-learning infrastructure for dramatically faster, cheaper, and lower-latency model deployment. You’ll work across GPU kernels, distributed serving, quantization, observability, and autoscaling in an on-site San Mateo environment.

Build polished, real-time product experiences for Moonlake’s AI-generated world simulations, blending frontend engineering, interaction design, and creative technology. You’ll own interfaces and supporting infrastructure for complex 3D, robotics, and embodied AI workflows.