Berk Çiçek

I recently completed my M.S. in Computer Engineering at Bilkent University, where I focused on Robot Learning, Vision-Language-Action Models, and Contact-Rich Manipulation. I was fortunate to be advised by Dr. Ozgur S. Oguz at the Bilkent Learning for Intelligent Robot Assistants Lab (LIRA).

My research bridges foundation models and embodied AI, with a particular focus on enabling robots to perform complex, contact-rich manipulation tasks through neuro-symbolic reasoning. My master’s thesis, Contact-VLA, introduces CoRAL (Contact-Rich Adaptive LLM-based Control)—a framework that leverages Large Language Models as adaptive world model designers and cost function synthesizers for model predictive control, achieving robust zero-shot manipulation on novel tasks.

My work has been published in venues including IEEE RA-L (with presentation at ICRA 2026) and ACS Photonics, with additional papers under review at RSS 2026. I’m passionate about developing AI systems that can reason about physical interactions, adapt to uncertainties, and generalize across diverse manipulation scenarios.

I obtained my B.S. in Mechanical Engineering from Bilkent University, where I developed a strong foundation in dynamics, control, and computational modeling that complements my work in robot learning.

Currently, I’m a Data Scientist at Trendyol developing AI-powered search algorithms, while continuing research collaborations at LIRA Lab. I’m also actively developing PuzzleWorld-3D, an API-based benchmark for evaluating embodied AI agents.

For more details, check out my CV, and Google Scholar.

Feel free to reach out via email—I’m always happy to discuss research ideas, collaborations, or just chat about robotics and AI!