Max Robotics — Global Service Robot Directory + US Certification

About

Diana Marculescu is a distinguished researcher whose work spans embedded systems education, efficient deep learning, and hardware-aware artificial intelligence. Her foundational contributions to undergraduate embedded systems curriculum at Carnegie Mellon University helped shape how an entire generation of engineers approaches the discipline, earning her 73 citations for that pioneering educational framework alone. More recently, Marculescu has positioned herself at the forefront of efficient AI and computer vision research. Her work on AdaScale demonstrated that speed and accuracy in video object detection need not be a fundamental trade-off — a significant insight for real-world autonomous systems like self-driving vehicles and robotics. Through her LeGR framework and related model compression research, she has advanced the field of neural network pruning by introducing learned global ranking strategies that remove the burden of manually specifying target model complexities, making deep learning more accessible and deployable on resource-constrained hardware. Her exploration of distributed reinforcement learning on CPU-GPU systems further reflects her commitment to bridging algorithmic innovation with architectural efficiency. Collectively, Marculescu's research addresses one of modern AI's most pressing challenges: making powerful machine learning models practical, fast, and energy-efficient for real-world deployment.