Constantine Caramanis (born August 25, 1977) is a professor in the Department of Electrical and Computer Engineering at the University of Texas at Austin, where he holds the Chandra Family Endowed Distinguished Professorship in Electrical and Computer Engineering. His research focuses on Machine learning, high-dimensional statistics, optimization, and decision-making in large-scale systems, with applications spanning communications, transportation, energy, and social networks.
Caramanis is known for contributions to robust and adaptable optimization, as well as the analysis of corrupted or incomplete data in high-dimensional settings. His work bridges theoretical foundations and practical algorithms, addressing challenges in Artificial intelligence and data science.
Education and career
Caramanis received an AB in mathematics from Harvard University. He then pursued graduate studies at the Massachusetts Institute of Technology, earning MS and PhD degrees in electrical engineering and computer science. His doctoral dissertation, completed in 2006, was titled Adaptable optimization: theory and algorithms.
In 2006, he joined the faculty of the University of Texas at Austin Department of Electrical and Computer Engineering. Over his career, he has supervised numerous doctoral students and collaborated with researchers across academic institutions and industry.
Research contributions
Caramanis's early work on adaptable optimization introduced methods for decision-making under uncertainty, where parameters are not fully known at the time of optimization. This line of research has applications in network design, logistics, and energy systems.
In high-dimensional statistics, he has studied problems involving corrupted or incomplete data, developing estimators and algorithms that remain reliable in the presence of outliers or missing entries. This work connects to Deep learning and Neural network theory, particularly in understanding how models behave with imperfect training data.
His recent research addresses Large language model efficiency and robustness, including topics related to Model Pruning and Data Augmentation. He has also explored optimization techniques such as stochastic gradient descent variants and learning rate schedules for training modern architectures.
Recognition and awards
In 2011, Caramanis received a National Science Foundation CAREER Award for research on high-dimensional statistics and the analysis of corrupted or incomplete data. This award supports early-career faculty who show potential to serve as academic role models.
He was elevated to Fellow of the IEEE in the 2023 class, an honor recognizing his contributions to robust statistics and high-dimensional optimization. The fellowship is reserved for individuals with outstanding accomplishments in IEEE-designated fields.
Caramanis has also served on program committees for major conferences in Machine learning and optimization, and has given invited talks at institutions including Berkeley and Oxford.
Teaching and service
At the University of Texas at Austin, Caramanis teaches courses on optimization, machine learning, and probability. He has mentored graduate students who have gone on to positions in academia and industry, including roles at Google DeepMind and OpenAI.
He has contributed to the broader research community through editorial service for journals and conference organization. His work often involves interdisciplinary collaboration, bridging electrical engineering, computer science, and statistics.
Selected publications
Caramanis has authored numerous papers in peer-reviewed journals and conference proceedings. Notable topics include robust principal component analysis, sparse recovery, and adaptive robust optimization. His publications are widely cited in the fields of high-dimensional statistics and optimization.
His research has been supported by federal agencies including the National Science Foundation and the Office of Naval Research, as well as industry partners. He continues to investigate fundamental questions in learning and decision-making under uncertainty.