Software Development, Research, Movies, and More

About Matt Trefilek

Originally from Illinois, I graduated in 2017 from Northern Michigan University. After graduation I began working as a full-stack developer for the University of Wisconsin - Madison. In 2021 I graduated from UW Madison with a Master's of Computer Science with a focus in Machine Learning and am working with Rentable (formerly ABODO) on interesting NLP and infrastructure work. I am passionate about computers and love expanding my personal knowledge and skills whenever I get the opportunity. Outside of computers, I am active in the outdoors, including hiking, biking, camping, skiing, and scuba diving.

Personal Projects

I am a full-stack and backend developer. I have large amounts of experience in Java, C++, and Javascript. Frameworks I have used include Spring Boot and Node.js. In my free time I enjoy working on a number of personal projects among many different technologies. Some of my work involves server administration using nginx, PM2, and Gitlab CI.


A collection of hosted webapps made primarily using node.js and using modules like express and


A Server/Client control for the homemade live action game built around the Garry's Mod gamemode known as Trouble in Terrorist Town. This game plays best in large spaces with lots of side rooms e.g. a museum. The rules can be found here.

Steam Secret Santa

A custom built Secret Santa platform for exchanging steam games with friends. Allows for custom date, price, and information settings. Includes email notifications and guaranteed to have to overlap in assignees.

Mobile Apps

Applications designed for Android and iPhone.

Swipe: The Game

Free game in which you swipe to remove dots in the correct direction. Features worldwide leaderboards, different difficulties, and fun gameplay!


As a hobby I run two different production "companies" where we make fun movies starring friends and family.


Here is a collection of research presentations and papers that I have worked on. I am passionate about Computer Science and love researching and expanding my knowledge.



Foundation Model Features are Strong Baseline for Private Learning (2021)

By utilizing large pre-trained models, we extract features for new datasets and can train a diferentially private model over the extracted feature vectors. By doing this, we propose that we can obfuscate individual sample information and still maintain a high accuracy for low ε values when compared to previous benchmarks.

Profiling Resource Bottlenecks in Sentiment Analysis (2019)

A final project written with a group for CS740 Big Data. using a sample Twitter Sentiment Dataset, we profiled the resource usage of cloud computing isntances as we scaled the classification problem.

A GAN Before Time (2019)

A fun final project written with a group for CS760 Machine Learning. This paper discusses a Generative Adversarial Network (GAN) set up to attempt to convert images of lizards into dinosaurs and vice versa.

Non-linear Dynamics in the A* Plane (2017)

The paper that goes with the presentation above. My senior project which focused on creating a client over which advanced simulations could be run using the A* Lattice, or a plane of Hexagons.

Configurations: The Complexity of Card Games and Chess (2015)

The final paper for MA340: Combinatorics. This paper focuses on the chessboard and offers equations that help get close to estimating the number of possible positions.

Implementation of Augmented Teleoperation (2015)

Created over the summer of 2015, this paper discusses my research on controlling robots using augmented reality. The research was conducted in part with the SULI program through the Argonne National Laboratory Nuclear Engineering Program.

Contact Me

Feel free to email or call me for just about anything!
(608) 247-5532