ABOUT ME

I am an Associate Research Scientist at the University of Michigan Transportation Research Institute (UMTRI) and an Affiliated Faculty member in both the Department of Robotics and the Michigan Institute for Data & AI in Society (MIDAS). My work sits at the convergence of Physical AI, Smart Cities, and Core Systems Engineering within complex cyber-physical environments. By bridging fundamental data science with real-world infrastructure deployment, my research program aims to transform autonomous systems from data-dependent models into fault-tolerant, mathematically bounded architectures capable of navigating real-world uncertainty.

Prior to my time at UMTRI, I spent five years as a Research Engineer at Ford Motor Company developing scalable autonomous vehicle (AV) pipelines. This unique blend of pioneering industrial R&D and high-impact academic leadership shapes my translational approach to engineering. Today, my lab manages a multi-million dollar external funding pipeline supported by agencies like the National Highway Traffic Safety Administration (NHTSA), Mity Integration Funds, and global corporate partners like Ford, Toyota, and Subaru.

Core Research Themes

• Infrastructure Perception & Domain Robustness: Diagnosing and mitigating the mathematical roots of context bias and out-of-distribution (OOD) anomalies in computer vision pipelines. Our lab develops lightweight, non-parametric statistical safety boundaries (such as Mask Pooling and Gaussian Process frameworks) to protect production-grade, roadside edge-computing units under extreme environmental variations without requiring post-hoc overhead.

• High-Fidelity Physics-Based Simulation: Quantifying and closing the Simulation-to-Reality ("Sim-to-Real") gap for V2X and automated driving systems. To bypass traditional computational bottlenecks, we engineer highly scalable, real-time LiDAR simulation architectures operating completely without raycasting, alongside generative models that simulate hardware degradation (e.g., failing sensors or fractured lenses) to stress-test smart networks prior to physical deployment.

• Topological Abstraction & Graph-Theoretical Edge Intelligence: Utilizing algebraic topology and graph theory to extract invariant structural "signatures" from raw sensory data. By mapping geometric invariant primitives onto high-dimensional hypervectors via Hyperdimensional (HD) Computing, we achieve real-time, low-power edge resilience against pixel-level corruptions, spatial occlusions, and severe weather shifts.

Innovation & Translational Impact

Driven by a commitment to translate academic discovery into market-ready solutions, I am the Co-founder and Chief Technology Officer of Proshno Labs LLC. Our mission is to commercialize robust, real-time sensor processing and uncertainty quantification solutions that streamline commercial infrastructure and perception pipelines.

My collaborative frameworks have received top-tier validation, including the Best Poster Award at AAAI-2026, selection for the ANRF Ramanujan Fellowship, leading the winning team for the TRB Data Forecasting Competition, and serving as the primary academic partner alongside Deloitte to win Stage 1A of the U.S. D.O.T. Intersection Safety Challenge.

I am actively looking to mentor graduate and undergraduate students who are passionate about defining the future of resilient, climate-adaptive smart cities. If you are interested in joining our smart infrastructure systems laboratory, please reach out via email.

Dr. Kusari was chosen as the academic partner to industry behemoths such as Deloitte, Nvidia, AWS etc, where he was tasked with grounding the possibilities of the solutions to be practical, flexible and above all, robust.