Ayça Takmaz

I am currently at Flexion, where I work on long-horizon autonomy for humanoid robots, building the spatial, embodied perception and reasoning abilities that let them carry out full missions on their own. Another core aspect I focus on is turning real-world captures and generated worlds into photorealistic, interactive world representations in which robots can train and be evaluated before acting in the physical world.

My PhD at ETH Zurich focused on how intelligent agents can understand and interact with complex 3D environments, not just recognizing what's in a scene, but reasoning about its geometry, semantics, and spatial relationships well enough to act in it, using language as a way to query and interact with these representations. I was supervised by Prof. Bob Sumner and Prof. Siyu Tang, working closely with Francis Engelmann. During my PhD, I also interned at Google with Johanna Wald and Federico Tombari, and at NVIDIA with Aljoša Ošep and Laura Leal-Taixé.

Prior to that, I did my Master's at ETH Zurich in computer vision and machine learning, working on research projects with Prof. Luc Van Gool and Prof. Marc Pollefeys.

I have also been co-organizing ZurichAI's Vision and Robotics branches since 2024, and the OpenSUN3D workshop series at ICCV, CVPR, and ECCV since 2023.