Vedhus Hoskere is an Associate Professor in the Department of Civil and Environmental Engineering at the University of California, Berkeley. Hoskere's research advances autonomous engineering across the infrastructure lifecycle, integrating engineering mechanics, artificial intelligence, and robotics to transform infrastructure design, construction, assessment, and maintenance. His group develops computer vision and spatial intelligence methods, physics-based synthetic environments and digital twins, engineering AI agents, and adaptive robotic systems for construction and repair. His research also integrates satellite observations and hazard information for rapid post-disaster assessment and recovery. His NSF CAREER project, BRACE-CI, advances open cyberinfrastructure for human-AI co-engineering and human-aligned master-builder agents to enable more resilient civil infrastructure. He develops increasingly capable engineering AI systems to augment human expertise and advance toward trustworthy civil engineering superintelligence.
Hoskere has led research supported by NSF, NASA, the U.S. Navy, the U.S. Army, and TxDOT. His honors include the 2026 NSF CAREER Award, the 2026 Cullen College Research Excellence Award, and best-paper recognition at ASCE EMI and i3CE conferences. He serves as an Associate Editor of the ASCE Journal of Computing in Civil Engineering. Before joining UC Berkeley, he served on the faculty at the University of Houston from 2020 to 2026, where he was the Kaspar J. Willam Associate Professor and directed the Structures and Artificial Intelligence Laboratory (SAIL).
- PhD, Civil Engineering, University of Illinois at Urbana-Champaign, 2020
- MS, Computer Science, University of Illinois at Urbana-Champaign, 2020
- MS, Civil Engineering, University of Illinois at Urbana-Champaign, 2016
- BS, Civil Engineering, BMS College of Engineering, India, 2014
Autonomous Infrastructure Lifecycle Engineering. Integrating artificial intelligence, engineering mechanics, and robotics into practical autonomous workflows spanning infrastructure design, construction, assessment, and maintenance (e.g., autonomous element-level bridge inspections, AI-assisted design, adaptive robotic 3D printing, and vision-guided crack repair).
Computer Vision & Artificial Spatial Intelligence. Advancing deep learning for fine-grained visual perception, 3D scene understanding, and spatial reasoning through foundation-model representation refinement, self-supervised learning, and language-guided change detection (e.g., PixelUp, EASe, and ViewDelta). Leveraging physics-based synthetic data and digital twins for training, evaluation, and generalization, providing foundations for physics-informed world models.
Engineering Agents, Superintelligence & Human-AI Alignment. Advancing engineering AI through synthetic datasets and testbeds (e.g., QuakeCity), evaluation benchmarks (e.g., BridgeEQA and PE Civil Bench), and hybrid generalist-specialist inspection systems. Evaluating reasoning, reliability, and alignment with engineering objectives. The NSF CAREER project BRACE-CI advances human-aligned master-builder agents and open cyberinfrastructure for human-AI co-engineering of more resilient civil infrastructure, supporting progress toward engineering superintelligence.
Construction Robotics & Adaptive Physical AI. Integrating deep learning-based perception, physics-based simulations of structural and material behavior, vision-language-action-force models and closed-loop control for robotic construction and repair (e.g., vision-guided crack repair, autonomous coating, and adaptive concrete 3D printing). Adapting robotic actions in real time to changing geometry and material behavior through feedback and continual learning.
Integrated Design for Autonomy. Combining structural design, material development, and robotic fabrication to enable autonomous construction and repair (e.g., machine-learning-guided material discovery, instant-curing composites, and robotic additive manufacturing).
Disaster Resilience & Autonomous Recovery. Leveraging multimodal sensing (e.g., optical and radar satellite imagery, aerial observations, and citizen science), physical hazard information (e.g., ground motion), and environmental and structural context for interpretable post-disaster damage assessment and recovery decisions.