BiographyI am a PhD student in the Department of Computer Science at Boston University. I am advised by Professor Jonathan Apavoo and Professor Steve Homer, and I often collaborate with Professor Orran Kreiger. I started my PhD in September 2017, and my main focus has been on operating systems and computer architecture. My current research interests range from operating systems structure and organization, efficiency and security in operating systems, and hybrid computing systems.
I received my BS in Computer Science from the American University of Beirut. During my undergraduate studies, I explored and attempted to integrate ideas from computer systems, machine learning, computer security, and neurobiology. One of my goals is to be able to incorporate learning and computation efficiently such that computation may be learned, stored, retrieved, and accelerated.
ResearchMy research interests mostly fall under systems research. Currently, I am reviewing the literature on different approaches to designing operating systems. Hence, I am studying the organization and structuring of operating systems including, but not limited to, monolithic operating systems, microkernel operating systems, library operating systems, and single address space operating systems.
The research direction that interests me most is that which incorporates systems research with machine learning and/or security. My first research collaboration was with Professor Jonathan Appavoo and Professor Steve Homer, during which I worked with the Programmable Smart Machines Lab on ASC (Automatically Scalable Computation). ASC is a model system that takes an alternative approach to automatic parallelization through tunnelling. It makes use of predictive partitioning of program trajectory and speculative execution of trajectory partitions, such that correct speculations incur speedup in program execution.
I am now collaborating with my colleagues and professors on designing Om, an operating system that maintains an executing target both as its classical OS construction and as a mathematical object. Om will allow us to study the execution trajectory of a target modeled mathematically while the target executes in OS world. Om will ultimately have the capacity to integrate the functionality of ASC within it in order to speedup program execution.
Some of my previous work was inspired by an intuition similar to that of programmable smart machines. I worked on predicting the number of context switches that a Linux target will undergo using target features extracted statically from the target's ELF. I also worked on enabling accurate malware detection and classification statically through modeling malware instances as document instances of a malware language.