Internship - Humanoid Motion Generation (Diffusion or Flow Matching)

Flexion is a fast‑growing robotics startup that creates the intelligence layer for next‑generation humanoid robots, aiming to move the technology from fragile prototypes to real‑world use. The company was founded by former Nvidia and ETH Zürich researchers and is backed by prominent venture capital firms. The internship is based in Zürich and focuses on advancing how humanoid robots perceive and act in three‑dimensional environments. As an intern you will explore generative approaches for whole‑body motion, experimenting with diffusion models, flow‑matching techniques, or auto‑regressive architectures while linking them to 3D perception pipelines. Your work will involve writing clean, modular Python code with PyTorch, rapidly prototyping new ideas, and integrating state‑of‑the‑art methods from recent research into the core software stack. The position requires you to be enrolled in a Master’s or PhD program in Computer Science, Robotics, Machine Learning, or a closely related discipline. You should have hands‑on experience with diffusion or flow‑matching models applied to trajectory generation, a solid grasp of generative modeling for time‑series data, and strong Python/PyTorch proficiency. Demonstrated analytical thinking, problem‑solving ability, and the capacity to implement cutting‑edge algorithms from the literature are essential. Prior work on human motion generation, robotic imitation learning, or ego‑centric action prediction is especially valued. Additional experience with robotics platforms, simulation tools, multi‑modal generative models, or fine‑tuning large foundation models such as Gr00t or SmolVLA would be advantageous. The role offers competitive pay, direct exposure to one of Europe’s most ambitious robotics ventures, and the chance to collaborate with an energetic, action‑oriented team.