Max Robotics — Global Service Robot Directory + US Certification

About

Valentin Radu is a computer systems researcher whose work sits at the intersection of deep learning and resource-constrained computing, with a particular focus on making convolutional neural networks (CNNs) practical for embedded and mobile deployment. His research addresses one of the most pressing challenges in modern AI: bridging the gap between the computational demands of state-of-the-art neural networks and the strict memory, energy, and processing limitations of edge devices. Radu's most influential contribution, "Characterising Across-Stack Optimisations for Deep Convolutional Neural Networks" (2018), has garnered over 40 citations and systematically investigates how optimisations spanning hardware, software, and algorithmic layers can be combined to enable efficient CNN inference. His follow-up work, TASO (2020), introduced a principled time-space co-optimization framework for memory-constrained inference, earning 12 citations and extending this vision further. His 2019 poster on integer linear programming for primitive selection demonstrated an elegant ahead-of-time approach to balancing competing resource constraints. Collectively, Radu's research provides both theoretical frameworks and practical tools for deploying intelligent systems in robotics, medical assistance, and biometric applications — making his work essential reading for engineers and researchers navigating the challenges of on-device AI.