John "Jack" Truskowski

Education

University of Wisconsin-Madison, Madison WI Sep. 2017 - Dec. 2019 (est)
M.S. Computer Science
Department of Computer Sciences

Bowdoin College, Brunswick ME Sep. 2013 - May 2017
B.A. Computer Science, minor in Music, Cum Laude

Work Experience

Teaching Assistant, University of Wisconsin-Madison Aug. 2017 - present

CS200 (4 semesters) - Introduction to Computer Science, a 600+ student course at UW-Madison taught in Java. Responsibilities include: Teach weekly lab sections, design and implement homework assignments, creation of an automated grading tool to grade hundreds of students' participation scores, grade student projects, hold weekly office hours.
CS300 (1 semester) - Programming II, taught in the summer on the campus of Epic Systems. Responsibilities include: Hold weekly office hours to help students on weekly programming assignments. Grading of exams. Creation of automated test suites to grade weekly programming assignments in Java. Proctor midterms and final exam.

Teaching Assistant, Bowdoin College Sep. 2014 - May 2017

Hold weekly office hours to assist students. Communicate regularly with professors to ensure students are receiving help and problem areas are addressed. Also employed as a grader and private tutor of problem sets and programming assignments for Data Structures, AI, and Introduction to Computer Science.

Cognitive Computing Intern, IBM Summer 2016

Worked in a team to apply machine learning to derive insights from customer data

Summer Research Fellow, Bowdoin College Summer 2015 - Fall 2016

Awarded a grant to design and implement an interactive music system in Max/MSP/Jitter. Backend written in C++.

Selected Papers and Code Samples

Grading Essays with Supervised Learning

(Abstract) In this paper, we apply various supervised learning techniques to the domain of automated essay grading. This task was originally examined in a 2012 Kaggle competition sponsored by the Hewlett Foundation. Our goal was to match or improve upon the results of the models from the competition, which we did: our best model would have ranked 32nd of the 154 competitors.

Transparent Gradient Aggregation in Machine Learning Systems

Used a Python wrapper to augment Horovod's architecture, enabling transparent gradient aggregation; if the compiler decides to revoke access to some subset of resources, an adequate aggregation strategy is chosen and work is re-distributed across machines. This enables restarting from the previous checkpoint without requiring input from the user, or a restart of the entire job.

Python Autograder for Class Participation

Written in Python, processes and automatically grades student participation on TopHat, a website for in-class participation. Used to grade hundreds of students across multiple semesters for Programming I at UW-Madison.

Jack Truskowski's Interactive Resume

Contact