AI Systems Researcher (Embedded Intelligence)

Your role is to push embodied intelligence toward human-level dexterity by working at the level where sensing, actuation, learning, and physical structure form a single closed loop. This role exists because dexterity is not a policy problem. It is a system problem. Intelligence in a humanoid does not live in a network alone—it emerges from how perception is structured, how actions are generated and constrained, and how the body itself participates in learning. Your work targets that loop directly. You work at the frontier of foundation models and multimodal learning, but you are not bound to existing architectures. You are expected to break with them when performance gaps demand it. You let failures in real systems—latency, instability, brittleness, lack of contact understanding—guide what models, representations, and interfaces need to exist next. Neuroscience and biological motor control are reference points, not inspiration slides. You explicitly embrace the difference between models built for language and models that must operate in tight sensor–actuator loops. You understand that embodiment imposes constraints—bandwidth, delays, noise, compliance—that fundamentally shape how intelligence must be structured. This role sits inside an interdisciplinary lab, embedded with hardware, sensing, biomechanics, and prototyping teams, while having direct access to 1X’s world-class AI organization. The loop between hypothesis, hardware change, experiment, and learning is intentionally short. You are expected to use that loop to unlock capabilities that cannot be reached by model-centric work alone.