MIT EECS Keel Foundation Undergraduate Research and Innovation Scholar
Charles E. Leiserson
Electrical Engineering and Computer Science
Performance Engineering in a Parallel Environment As Moore's law ends and compute-intensive processes such as machine learning becomes mainstream we must look to parallelism to allow our technological dreams to progress. Historically parallelism has been implemented with disjoint building blocks hacked together to make the whole thing run. The net result is weak parallel performance "€œ especially compared with serial code! The goal of this project is to holistically integrate parallelism into a programmer's stack: modify parallel linguistics to make them both more optimizable and easier to use a compiler that understands parallel code to perform clever parallel optimizations on it and a low-level runtime system that is able to fully take advantage of the hardware below it.
I'm excited to be able to be at the forefront of parallel programming and see what sorts of linguistic and performance improvements we can get. I've been interested in parallelism after developing a high-level language for running physics simulations a few years back. Since then I've done research on creating both fast compilers and neural networks and I'm excited to see where this takes me.