Senior Research Scientist, Google DeepMind · Robotics
Hi! I’m Cem (pronounced Jem — Turkish phonetics!). I’m a Senior Research Scientist at Google DeepMind, working on robotics simulation — on a quest to make robots boring.
I build the simulators, benchmarks and learning algorithms that let robots perform complex, long-horizon tasks in real homes, drawing on learning from demonstrations, reinforcement learning, sim-to-real transfer, and LLMs/VLMs. I co-developed and maintained OmniGibson (built on NVIDIA Omniverse / Isaac Sim) and iGibson, and led the team that built the BEHAVIOR-1K asset dataset.
I did my Ph.D. in Computer Science at Stanford in the Stanford Vision & Learning Lab, advised by Prof. Fei-Fei Li. Along the way I interned at Meta FAIR, NVIDIA Research and Google [X] / Everyday Robots, and before that was a software engineer at Google (YouTube). BS in CS from Georgia Tech.
News
Joined Google DeepMind as a Senior Research Scientist, continuing my work on robotics simulation.
Completed my Ph.D. in Computer Science at Stanford — read the thesis.
Started a research internship at Meta FAIR on generating robot simulation scenes from house tour videos.
MoMaGen accepted to ICLR 2026.
BEHAVIOR Robot Suite accepted to CoRL 2025.
Embodied Agent Interface selected as an Oral at NeurIPS 2024 Datasets & Benchmarks.
ACDC (Digital Cousins) accepted to CoRL 2024.
Head CA for CS231N (Spring 2024).
BEHAVIOR Vision Suite is a Highlight at CVPR 2024.
Interned at NVIDIA Research with Jim Fan & Yuke Zhu.
“Asking for Help” (BCVA) at ICRA 2023.
BEHAVIOR-1K — Oral & Best Paper Nominee at CoRL 2022.
AI Resident at Google [X] / Everyday Robots.
Research
Selected papers. All publications ↓
-
A real-world whole-body manipulation stack — bimanual wheeled robot, teleoperation interface, and visuomotor policy learning — for household tasks.
-
Automatically generated "digital cousins" — simulated scenes with similar geometry and semantics to the real world — transfer zero-shot at 90% success vs. 25% for digital twins.
-
Embodied Agent Interface: Benchmarking LLMs for Embodied Decision Making
- NeurIPS 2024 D&B
- Oral
A standardized, fine-grained framework to benchmark LLMs for embodied decision making, evaluating 15 LLMs on BEHAVIOR and VirtualHome.
-
BEHAVIOR Vision Suite: Customizable Dataset Generation via Simulation
- CVPR 2024
- Highlight
Fully customizable synthetic data generation (scene, object, camera level) for systematically evaluating computer vision models.
-
BCVA predicts failures of imitation-learned policies so robots can ask for help — 86% precision across 2,000+ real-world trials at Everyday Robots.
- arXiv
- First author
-
BEHAVIOR-1K: A Benchmark for Embodied AI with 1,000 Everyday Activities and Realistic Simulation
- CoRL 2022
- Oral
- Best Paper Nominee
A human-centered embodied AI benchmark of 1,000 everyday activities, with a 6k-object / 50-scene dataset and the OmniGibson simulator.
- Website
- arXiv
- Code
- Knowledgebase
- Co-first author
All publications
* equal contribution
-
ICLR 2026
-
CoRL 2025
-
ACDC: Automated Creation of Digital Cousins for Robust Policy Learning
CoRL 2024
-
Embodied Agent Interface: Benchmarking LLMs for Embodied Decision Making
NeurIPS 2024 Datasets & Benchmarks Track Oral
-
BEHAVIOR Vision Suite: Customizable Dataset Generation via Simulation
CVPR 2024 Highlight
-
Asking for Help: Failure Prediction in Behavioral Cloning through Value Approximation
ICRA 2023
-
BEHAVIOR-1K: A Benchmark for Embodied AI with 1,000 Everyday Activities and Realistic Simulation
CoRL 2022 Oral Best Paper Nominee
-
Eye-BEHAVIOR: An Eye-Tracking Dataset for Everyday Household Activities in Virtual, Interactive, and Ecological Environments
Journal of Vision, 2022
-
CoRL 2021
-
iGibson 2.0: Object-Centric Simulation for Robot Learning of Everyday Household Tasks
CoRL 2021
-
A Local Stochastic Algorithm for Separation in Heterogeneous Self-Organizing Particle Systems
RANDOM 2019
Experience
-
Google DeepMind Robotics
Senior Research Scientist · Robotics simulation
Jun 2026 – present
-
Meta FAIR
AI Research Scientist Intern · Generating robot simulation scenes from in-the-wild house tour videos
Jun 2025 – Jan 2026
-
Stanford Vision & Learning Lab
Graduate Research Assistant · Advised by Prof. Fei-Fei Li
Jan 2021 – Jun 2026
-
NVIDIA Research GEAR
Research Scientist Intern · LLMs for robot planning on long-horizon household activities · Hosts: Jim Fan, Yuke Zhu
Jan – Dec 2023
-
Google [X] / Everyday Robots
AI Resident · Learning from demonstrations · Hosts: Mohi Khansari, Daniel Ho
Jun – Sep 2022
-
Google YouTube Premium
Software Engineer
Feb 2019 – Sep 2020
-
Georgia Tech Algorithms & Randomness Center
Undergraduate Researcher · Randomized algorithms, advised by Prof. Dana Randall
2017 – 2018
Education
-
Stanford University
Ph.D., Computer Science · Advisor: Prof. Fei-Fei Li · Thesis
2022 – 2026
-
Stanford University
M.S., Computer Science · Google CSRMP Fellow
2020 – 2022
-
Georgia Institute of Technology
B.S., Computer Science · AI & Theory tracks · Undergraduate Research Certification
2016 – 2018
Teaching
-
CS231N: Deep Learning for Computer Vision Stanford
Head Course Assistant · Led 22 CAs for a 650-student course
Spring 2024
-
CS107: Computer Organization & Systems Stanford
Course Assistant
2020 – 2021
-
CS2110: Computer Organization & Programming Georgia Tech
Senior Teaching Assistant · 96% effectiveness rating; designed course materials for 400+ students
2017 – 2018