Feras Saad

I recently completed the PhD in EECS at MIT (MEng/SB 2016), where I worked with Vikash Mansinghka at the Probabilistic Computing Project and Martin Rinard at the Computer Science and Artificial Intelligence Laboratory. I work broadly in the areas of programming languages, statistics, and artificial intelligence.

I am pleased to be joining the Computer Science Department at Carnegie Mellon University as an Assistant Professor in Fall 2023. Before joining, I am spending one year as a Visiting Research Scientist at Google. My group is recruiting students and postdocs. If you are interested in working with me, please send me an email to fsaad@cmu.edu and (prospective students) apply to the CMU CS PhD program.


I am interested in developing techniques that enable large-scale probabilistic modeling, inference, and computation across challenging application domains. Some current research themes include:

Probabilistic programming. Programs give us a uniquely expressive formalism for modeling and understanding complex empirical phenomena. My work develops new programmable systems that help automate, formalize, and scale-up the very hard aspects of modeling and inference. [PLDI-21] [POPL-19] [PLDI-19] [AISTATS-17] [NIPS-16]

Automatically discovering models from data. A grand challenge of AI is the ability to automate the process of discovering accurate and interpretable models from data. A mathematically elegant and practical approach to this problem is Bayesian structure learning over rich probabilistic model families. [ICML-23] [UAI-21] [POPL-19] [AISTATS-18]

Simulation-based statistical estimators and tests. As probabilistic programs become more widespread for representing complex probability distributions, scalable simulation-based techniques are needed to analyze their statistical properties using the black-box computational interfaces they expose. [AISTATS-22] [AISTATS-19]

Fast random sampling algorithms. This thread explores fundamental computational limits of random sampling; including new algorithms that are theoretically optimal/near-optimal (in entropy and error) and extremely efficient in practice. [POPL-20] [AISTATS-20].

Software and Applications. A central goal of my work is to build performant, freely available software systems that make probabilistic modeling and inference more broadly accessible and help domain-specialists solve high-impact applied problems in the sciences, engineering, and public interest. [eLife-2019]



Software and repositories from research projects (2500+ Github stars).

Sublime Text Users: You may be interested in these productivity plugins (8000+ users):
AddRemoveFolder; RemoveLineBreaks; ViewSetting.