Sanjar Atamuradov
Founder and robotics researcher focused on humanoid teleoperation
San Francisco, California
Summary
Founder building teleoperation-first infrastructure to scale robot learning datasets: At Humanola he focuses on an end-to-end telepresence platform that enables VR-based remote control of robots while automatically collecting, cleaning and annotating multimodal session data so those interactions can be reused as training data for robot learning. gatech+2
Researcher and author in robot learning and teleoperation: He has authored arXiv papers on neural teleoperation and meta-learning for robotic manipulation and maintains academic profiles (Google Scholar, Semantic Scholar) tied to his Georgia Tech affiliation. arxiv+3
Hands-on industry experience across diverse robotic platforms: His background spans computer vision, AMRs for warehouse automation, quadruped locomotion, and mobile robots for security/inspection—bridging applied engineering and academic research. it-park
International academic and professional trajectory: From KAIST research labs (AVE, IRiS) in South Korea to graduate work and research at Georgia Tech in the U.S., he combines cross-cultural education and lab experience with Silicon Valley entrepreneurship. it-park+1
Work
Education
Projects
Writing
Learning Adaptive Neural Teleoperation for Humanoid Robots: From Inverse Kinematics to End-to-End Control
November 1, 2025Proposes a learning-based neural teleoperation framework that maps VR controller inputs directly to robot joint commands, trained via reinforcement learning to improve tracking, smoothness and force adaptation over traditional IK+PD pipelines; validated in simulation and on humanoid hardware.
Evaluating Model-Agnostic Meta-Learning on MetaWorld ML10 Benchmark: Fast Adaptation in Robotic Manipulation Tasks
November 1, 2025Evaluates MAML combined with TRPO on the MetaWorld ML10 benchmark, analyzing rapid adaptation capabilities and generalization gaps for diverse robotic manipulation skills.