Research Engineer - Generative Humanoid Motion Generation

Flexion is creating the intelligence layer that will enable humanoid robots to move from fragile prototypes to real‑world use. The Zurich‑based team, founded by experts in robot reinforcement learning, is seeking a Research Engineer who will focus on whole‑body motion generation and vision‑conditioned generative models. The day‑to‑day work involves designing, training, and fine‑tuning cutting‑edge multimodal generative models—such as diffusion, flow‑matching, or masked‑Transformer architectures—to produce robot trajectories. The engineer will also integrate these models into robotic systems, run GPU‑accelerated simulations using platforms like Omniverse or Genesis, and collaborate closely with colleagues to bring the generated motions to the lab’s testing facilities. Candidates need a PhD or master’s in generative modeling, human motion generation, or learning‑based robot trajectory generation, plus a strong publication record at venues such as ICCV, CVPR, SIGGRAPH, CORL or RSS. Proficiency in Python and PyTorch, experience with training and deploying generative models, and familiarity with modern simulation tools are required. Helpful but optional skills include multi‑modal generation, ego‑centric action prediction, and finetuning foundation models. The role offers competitive pay, enhanced pension and holiday benefits, relocation and permit support, and a vibrant, fast‑moving environment in central Zürich with top‑tier robotics infrastructure.