About Me

I'm a first-year graduate student at Stanford University. I'm studying computer science, interested in systems research, and am currently working with with Peter Bailis and Matei Zaharia on the MacroBase project. Before Stanford, I did my undergrad at Harvard, where I majored in computer science and worked with Professor Margo Seltzer on the Automatically Scalable Computation (ASC) project, writing my senior thesis on my work on it. I also worked with Professor Alexander Rush on a natural language processing project of predicting congressional voting records from bill text, which made it into EMNLP 2016.

Contact Details

Peter Kraft
kraftp@cs.stanford.edu

Education

Harvard University

Bachelor's in Computer Science 2017

I concentrated in Computer Science at Harvard, graduating in 2017. My undergrad and thesis advisor was Professor Margo Seltzer. I ended up taking and, later, being a teaching fellow for the notorious CS 161.

I worked with Professor Seltzer on the Automatically Scalable Computation (ASC) project. The aim of the project was to develop a system that automatically parallelized sequential programs by predicting future behavior of the program and then speculating from the predictions, speeding up the program with the results of the speculations if the predictions turned out to be accurate. My work substantially improved the efficacy and scalability of ASC, demonstrating real speedup on a variety of programs. This project became my senior thesis, which received highest honors and won the Hoopes Prize for excellence in undergraduate research.

I also worked with Professor Alexander Rush and my friend Hirsh Jain on a natural language processing project predicting the results of congressional votes based purely on the text of the bills voted on. We developed a novel approach to the problem, creating a powerful and simple model that both predicted votes with over 90% accuracy (higher than any previous approach) and easily lent itself to analysis, offering useful insights into the behavior of congresspeople. This work was published in EMNLP 2016.

Stanford University

Ph.D. in Computer Science Eventually

I'm in my first year at Stanford and am advised by Peter Bailis and Matei Zaharia working on Macrobase. I'm hoping to continue doing systems research here and am excited for what I can accomplish in the future!

Publications

At Harvard

Peter Kraft, Hirsh Jain, Alexander Rush, "An Embedding Model For Predicting Roll-Call Votes." Empirical Methods for Natural Language Processing 2016. Link.

Peter Kraft, "Automatically Scalable Computation That Is More Scalable and Automatic." Senior Thesis. Harvard University. Link.

At Stanford

Coming soon!