May 27, 2025
EECS Professor and LIDS PI Devavrat Shah has received the 2025 ACM SIGMETRICS Achievement Award. ACM’s announcement cites Shah’s “contributions to the performance analysis and design of computer and communication networks.”
Shah is the Andrew and Erna Viterbi Professor of Electrical Engineering and Computer Science at MIT. His research focuses on statistical inference and stochastic networks. His contributions span a variety of areas including distributed algorithms for computer communications networks, inference and learning on graphical models, algorithms for social data processing including ranking, recommendations and crowdsourcing and, more recently, causal inference.
In the area of distributed algorithms, Shah has made several key advancements: In the space of Medium Access Protocols, he developed a network-agnostic algorithm that delivers maximum throughput; Gossip Algorithms, where he analyzed the convergence time for evaluating global and separable functions through local node communications; Belief Propagation, in which he connected algorithmic performance to classical network flow optimization; and he has developed a PubSub-based architecture for message-passing algorithms. Additionally, his work on Patient-Zero identification has established statistical methods for detecting the origin of a contagion. He has introduced time series algorithms through connecting the statistical methods with low-rank matrix and tensor estimation.
Shah’s developments in the areas of social choice and ranking systems include: The development of the “Rank Centrality Algorithm” for efficiently creating global rankings from sparse pairwise comparisons; the establishment of frameworks for learning high-dimensional distributions over rankings from partial preferences; and providing theoretical foundations for Collaborative Filtering in recommendation systems.
More recently, he has advanced causal inference methodologies, particularly for data-rich environments. He has bridged structural causal models with latent factor models to address unobserved confounding and to enable consistent estimation of causal effects. This has enabled unit-level counterfactual inference for personalized therapeutics in Alzheimer’s Disease and Oncology, societal policy evaluation, and trace-driven simulation for communication network protocols. Viewed together, these contributions demonstrate Shah's robust and practical toolkit for drawing causal conclusions in a variety of complex, real-world scenarios.
His advancements have been recognized with several awards, including The ACM Sigmetrics Test of Time Awards (2019 and 2020), and best paper and student paper awards at ACM Sigmetrics (2006, 2009), NeurIPS (2008), IEEE INFOCOM (2004), INFORMS Applied Probability Society (2012), the INFORMS Revenue Management and Pricing (2015) and the INFORMS Management Science and Operations Management (2016). He has received career recognition awards; notably, the 2008 ACM Sigmetrics Rising Star Award, the 2010 Erlang Prize from INFORMS Applied Probability Society and he delivered the 2024 INFORMS Applied Probability Society Markov Lecture. He is an IEEE Fellow and a Kavli Fellow of the U.S. National Academy of Science. He is a distinguished alumnus of his alma mater IIT Bombay.
In addition to academic significance, these contributions have had substantial industry impact leading to the founding of Celect (later acquired by Nike) which helped retailers optimize inventory, and founding of Ikigai Labs to enable large enterprises to transform their forecasting and planning.
SIGMETRICS is the ACM Special Interest Group (SIG) for the computer performance evaluation community. SIGMETRICS promotes research in performance analysis techniques as well as the advanced and innovative use of known methods and tools. It sponsors conferences, such as its own annual conference (SIGMETRICS), whose papers appear in the Proceedings of the ACM on Measurement and Analysis of Computer Systems (POMACS). It also publishes a newsletter (Performance Evaluation Review), and operates a mailing list linking researchers, students, and practitioners interested in performance evaluation.